ResnetInceptionV2_Oversampling.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "_PrjgC87ZbNT"
   },
   "outputs": [],
   "source": [
    "#install depenencies\n",
    "!pip install kaggle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 92,
     "resources": {
      "http://localhost:8080/nbextensions/google.colab/files.js": {
       "data": "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",
       "headers": [
        [
         "content-type",
         "application/javascript"
        ]
       ],
       "ok": true,
       "status": 200,
       "status_text": ""
      }
     }
    },
    "colab_type": "code",
    "id": "LYiwHG4dZuSs",
    "outputId": "200807fc-a05f-4b78-e35b-3ad246951c75"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "     <input type=\"file\" id=\"files-6f6ab1a5-ed3d-4864-8d5d-3627ec1d5efa\" name=\"files[]\" multiple disabled />\n",
       "     <output id=\"result-6f6ab1a5-ed3d-4864-8d5d-3627ec1d5efa\">\n",
       "      Upload widget is only available when the cell has been executed in the\n",
       "      current browser session. Please rerun this cell to enable.\n",
       "      </output>\n",
       "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving kaggle.json to kaggle.json\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'kaggle.json': b'{\"username\":\"siddharthnaik7\",\"key\":\"28e0f6c07a924550d841c251e1730ac7\"}'}"
      ]
     },
     "execution_count": 2,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#upload the credentials of the kaggle account\n",
    "from google.colab import files\n",
    "files.upload()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "7HPubmLlaW9U"
   },
   "outputs": [],
   "source": [
    "#before importing the dataset we want to use this code\n",
    "# The Kaggle API client expects this file to be in ~/.kaggle,\n",
    "!mkdir -p ~/.kaggle\n",
    "!cp kaggle.json ~/.kaggle/\n",
    "\n",
    "# This permissions change avoids a warning on Kaggle tool startup.\n",
    "!chmod 600 ~/.kaggle/kaggle.json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "P31EWwjeagb6"
   },
   "outputs": [],
   "source": [
    "#import the dataset we want to use for our project\n",
    "!kaggle competitions download -c aptos2019-blindness-detection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 80
    },
    "colab_type": "code",
    "id": "RGuo0kezccd3",
    "outputId": "b2759cf0-771d-4be3-8e74-6b519bfa5a91"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "kaggle.json  sample_submission.csv  test_images.zip  train_images.zip\n",
      "sample_data  test.csv\t\t    train.csv\n"
     ]
    }
   ],
   "source": [
    "#lets see the imported files from kaggle competetion\n",
    "!ls"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Bv_FyDqg373l"
   },
   "source": [
    "# **Unzip Data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "zKrcCu8Mgd0L"
   },
   "outputs": [],
   "source": [
    "!unzip train_images.zip -d ./Train_Images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Eycej2e2g-Vk"
   },
   "outputs": [],
   "source": [
    "!unzip test_images.zip -d ./Test_Images"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "bQ_QtMdF4C5T"
   },
   "source": [
    "# **Delete Zip Files**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "lnVZdHqao57J"
   },
   "outputs": [],
   "source": [
    "!rm -rf test_images.zip"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "0C6TDe-GpCyW"
   },
   "outputs": [],
   "source": [
    "!rm -rf train_images.zip"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Iw5FHIGb4I8j"
   },
   "source": [
    "# **Load Training and Test Information in Dataframes**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "Rd4hUOAPpIEJ",
    "outputId": "e9904683-183d-498f-c50f-08238557e906"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id_code</th>\n",
       "      <th>diagnosis</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>000c1434d8d7</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>001639a390f0</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0024cdab0c1e</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>002c21358ce6</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>005b95c28852</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        id_code  diagnosis\n",
       "0  000c1434d8d7          2\n",
       "1  001639a390f0          4\n",
       "2  0024cdab0c1e          1\n",
       "3  002c21358ce6          0\n",
       "4  005b95c28852          0"
      ]
     },
     "execution_count": 286,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "df_train = pd.read_csv('train.csv')\n",
    "df_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "57-4ClMspYgK",
    "outputId": "9e829aa2-6ea9-4e3f-c196-715b80638c33"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id_code</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0005cfc8afb6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>003f0afdcd15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>006efc72b638</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>00836aaacf06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>009245722fa4</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        id_code\n",
       "0  0005cfc8afb6\n",
       "1  003f0afdcd15\n",
       "2  006efc72b638\n",
       "3  00836aaacf06\n",
       "4  009245722fa4"
      ]
     },
     "execution_count": 287,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test = pd.read_csv('test.csv')\n",
    "df_test.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "kT2LbBF12zk3"
   },
   "source": [
    "# **Base Line**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 420
    },
    "colab_type": "code",
    "id": "RaibCVoi22_Z",
    "outputId": "d6d51f49-ed5c-4612-fb4d-4fdb1a3583f7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0    1805\n",
      "2     999\n",
      "1     370\n",
      "4     295\n",
      "3     193\n",
      "Name: diagnosis, dtype: int64\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZkAAAEcCAYAAAAV2MmlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nZMOQISIi2TBkiIhINv8PHgQ3qvD3TAYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "Class_Frequency_Count = df_train['diagnosis'].value_counts()\n",
    "print(str(Class_Frequency_Count) + \"\\n\")\n",
    "sns.set(style=\"darkgrid\")\n",
    "sns.barplot(Class_Frequency_Count.index,Class_Frequency_Count.values,alpha=0.9)\n",
    "plt.title(\"Frequency Distribution of Classes\")\n",
    "plt.ylabel(\"Number of Occurances\", fontsize=12)\n",
    "#plt.xticks(rotation=90)\n",
    "plt.xlabel(\"Classes\", fontsize=12)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "M6edlOHG3ys4"
   },
   "source": [
    "# **Directory Structure Creation**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "y2gncbKq0gaE"
   },
   "outputs": [],
   "source": [
    "!rm -rf Training_Dir\n",
    "!rm -rf Validation_Dir"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "UYlNiuj6--NC"
   },
   "outputs": [],
   "source": [
    "df_train = df_train.sample(frac=1).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "q8oCp3uMrnfC"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "train_keyword = 'Training_Dir'\n",
    "validation_keyword = 'Validation_Dir'\n",
    "keywords = [train_keyword,validation_keyword]\n",
    "dir_list = df_train['diagnosis'].tolist()\n",
    "\n",
    "for keyword in keywords:\n",
    "  if(not os.path.exists(keyword)):\n",
    "    os.mkdir(keyword)\n",
    "  for class_dir in dir_list:\n",
    "    if(not os.path.exists(keyword+'/'+str(class_dir))):\n",
    "      os.mkdir(keyword+'/'+str(class_dir))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 289
    },
    "colab_type": "code",
    "id": "enj-3e567QHr",
    "outputId": "aff5b2fa-b2b1-48b6-8d39-4e5a44bdcb75"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:2: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  \n",
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:3: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "X = np.array(df_train.ix[:, df_train.columns == 'id_code'])\n",
    "y = np.array(df_train.ix[:, df_train.columns == 'diagnosis'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "JAVXFt8TJPuv"
   },
   "outputs": [],
   "source": [
    "la =[]\n",
    "count = 0\n",
    "for x in X:\n",
    "  if x in la:\n",
    "    print('same')\n",
    "  else:\n",
    "    la.append(count)\n",
    "    count=count+1\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nEzFkQUYKQvl"
   },
   "outputs": [],
   "source": [
    "X_update = np.array(la)\n",
    "X_update =X_update.reshape(-1, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "jGn9ip51tx9L"
   },
   "outputs": [],
   "source": [
    "re = dict(zip(la,df_train['id_code'].tolist()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "L49lDTFq7abE"
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split #  train,test = train_test_split(dataset,test_size=0.20,stratify=True)\n",
    "train_images_temp, val_images, train_labels_temp, val_labels = train_test_split(X_update, y, test_size=0.2,stratify=y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nXbM4Q0xpj7x"
   },
   "outputs": [],
   "source": [
    "#NU\n",
    "split = int(0.7 * len(df_train)) # Change V2 0.8 to 0.7\n",
    "\n",
    "train_dataframe = df_train.iloc[:split]\n",
    "val_dataframe = df_train.iloc[split:]\n",
    "\n",
    "#train_images = train_dataframe['id_code']\n",
    "#train_labels = train_dataframe['diagnosis']\n",
    "val_images = val_dataframe['id_code'].tolist()\n",
    "val_labels = val_dataframe['diagnosis'].tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 306
    },
    "colab_type": "code",
    "id": "QYVlzkvfYoDa",
    "outputId": "0cd2d737-ef09-4a42-b3a8-697d3ff072c0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before OverSampling, counts of label '0': [1444]\n",
      "Before OverSampling, counts of label '1': [296] \n",
      "\n",
      "Before OverSampling, counts of label '2': [799]\n",
      "Before OverSampling, counts of label '3': [154] \n",
      "\n",
      "Before OverSampling, counts of label '4': [236]\n",
      "After OverSampling, the shape of train_X: (4219, 1)\n",
      "After OverSampling, the shape of train_y: (4219,) \n",
      "\n",
      "Before OverSampling, counts of label '0': 1444\n",
      "Before OverSampling, counts of label '1': 296 \n",
      "\n",
      "Before OverSampling, counts of label '2': 799\n",
      "Before OverSampling, counts of label '3': 1444 \n",
      "\n",
      "Before OverSampling, counts of label '4': 236\n"
     ]
    }
   ],
   "source": [
    "from imblearn.over_sampling import SMOTE\n",
    "print(\"Before OverSampling, counts of label '0': {}\".format(sum(train_labels_temp==0)))\n",
    "print(\"Before OverSampling, counts of label '1': {} \\n\".format(sum(train_labels_temp==1)))\n",
    "print(\"Before OverSampling, counts of label '2': {}\".format(sum(train_labels_temp==2)))\n",
    "print(\"Before OverSampling, counts of label '3': {} \\n\".format(sum(train_labels_temp==3)))\n",
    "print(\"Before OverSampling, counts of label '4': {}\".format(sum(train_labels_temp==4)))\n",
    "\n",
    "sm = SMOTE('minority')\n",
    "train_images, train_labels = sm.fit_sample(train_images_temp, train_labels_temp.ravel())\n",
    "\n",
    "print('After OverSampling, the shape of train_X: {}'.format(train_images.shape))\n",
    "print('After OverSampling, the shape of train_y: {} \\n'.format(train_labels.shape))\n",
    "\n",
    "print(\"Before OverSampling, counts of label '0': {}\".format(sum(train_labels==0)))\n",
    "print(\"Before OverSampling, counts of label '1': {} \\n\".format(sum(train_labels==1)))\n",
    "print(\"Before OverSampling, counts of label '2': {}\".format(sum(train_labels==2)))\n",
    "print(\"Before OverSampling, counts of label '3': {} \\n\".format(sum(train_labels==3)))\n",
    "print(\"Before OverSampling, counts of label '4': {}\".format(sum(train_labels==4)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "rsUf4DRHObkq"
   },
   "outputs": [],
   "source": [
    "train_temp=[]\n",
    "for ds in train_images:\n",
    "  train_temp.append(re[ds[0]])\n",
    "train_images = train_temp\n",
    "train_labels = train_labels.tolist()\n",
    "\n",
    "val_temp=[]\n",
    "for ds in val_images:\n",
    "  val_temp.append(re[ds[0]])\n",
    "val_images = val_temp\n",
    "val_labels = val_labels.tolist()\n",
    "cx = []\n",
    "for l in val_labels:\n",
    "  cx.append(l[0])\n",
    "val_labels = cx"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "h4ViD1Hc7thA"
   },
   "outputs": [],
   "source": [
    "list_of_tuples = list(zip(train_images,train_labels))\n",
    "train_dataframe = pd.DataFrame(list_of_tuples,columns = ['id_code', 'diagnosis'])\n",
    "train_dataframe = train_dataframe.sample(frac=1).reset_index(drop=True)\n",
    "train_images = train_dataframe['id_code'].tolist()\n",
    "train_labels = train_dataframe['diagnosis'].tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "xICWy9TUHkTE"
   },
   "outputs": [],
   "source": [
    "list_of_tuples = list(zip(val_images,val_labels))\n",
    "val_dataframe = pd.DataFrame(list_of_tuples,columns = ['id_code', 'diagnosis'])\n",
    "val_dataframe = val_dataframe.sample(frac=1).reset_index(drop=True)\n",
    "val_images = val_dataframe['id_code'].tolist()\n",
    "val_labels = val_dataframe['diagnosis'].tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 420
    },
    "colab_type": "code",
    "id": "Nf2EgnLMgwgw",
    "outputId": "637ddc62-b3d7-4164-bb6a-180a4dd0d87b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3    1444\n",
      "0    1444\n",
      "2     799\n",
      "1     296\n",
      "4     236\n",
      "Name: diagnosis, dtype: int64\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZkAAAEcCAYAAAAV2MmlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkWz+D2WuxmECzN/IAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "Class_Frequency_Count = train_dataframe['diagnosis'].value_counts()\n",
    "print(str(Class_Frequency_Count) + \"\\n\")\n",
    "sns.set(style=\"darkgrid\")\n",
    "sns.barplot(Class_Frequency_Count.index,Class_Frequency_Count.values,alpha=0.9)\n",
    "plt.title(\"Frequency Distribution of Classes\")\n",
    "plt.ylabel(\"Number of Occurances\", fontsize=12)\n",
    "#plt.xticks(rotation=90)\n",
    "plt.xlabel(\"Classes\", fontsize=12)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 420
    },
    "colab_type": "code",
    "id": "b9Y-pdIGhGwJ",
    "outputId": "978bf71c-124f-45bf-aad1-15a2dea860a1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0    361\n",
      "2    200\n",
      "1     74\n",
      "4     59\n",
      "3     39\n",
      "Name: diagnosis, dtype: int64\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZIAAAEcCAYAAADtODJSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzt3XlcVPX+P/DXDDAou5DAoJbmTSG5\nXtFR8iqlqLmESoVLaJZL5gKhhqFpYCgiYOZyMTKXlmt601LU3MqtMjP9lilXLTMlhWEdUHZw5vP7\nw59zJRYPDsPM4Ov5ePR4MJ/PWd5zGud1PuecOUcmhBAgIiK6T3JTF0BERJaNQUJERAZhkBARkUEY\nJEREZBAGCRERGYRBQkREBmGQENViypQp2LFjR6Ms6/Tp0xg8eLD+dWBgIL7//vtGWTYAPPPMMzh5\n8mSjLU8KIQTmz5+Pnj17IiQk5L6W0blzZ6SnpzdyZWQK1qYugJpGYGAg8vLyYGVlpW/bv38/PDw8\nTFiVaXTu3BktW7aETCaDQqGAt7c3xowZg2HDhumnWb9+veRlHTx4EI888kid06hUKhw4cMDgugFg\n3rx58PDwwOzZs/VtX375ZaMsuyH+7//+D8ePH8exY8dgZ2dX6zQ5OTlYuXIlvvnmG5SUlMDDwwPD\nhg3DlClT6pyHLBOD5AGSkpKCf/7zn/VOc+vWLVhbN/+PRWpqKh555BFoNBp88803iI2NxR9//IGw\nsLBGXU9z3Z4ZGRlo06ZNnYFQWFiIsWPHws/PD1u3bkXbtm2hVquxYcMG/Pnnn/D29m7iismYeGjr\nAXf9+nV07twZ27ZtQ79+/fDSSy8BAM6cOYOxY8dCpVJhxIgR1Q6dXLt2DePHj4efnx8mTpyI2NhY\nREZGAgBOnjyJJ598sto67j6Uo9PpsG7dOgwcOBD+/v6IiIhAYWFhtVp27NiBfv36wd/fH++9955+\nOVqtFikpKRg4cCD8/Pzw3HPPQa1W4+2338ayZcuqrXPatGn48MMP7/n+XV1dERwcjEWLFuH9999H\nQUEBAODFF1/Etm3bAADp6ekYP348evToAX9/f8yaNQsAMG7cOADAyJEj4efnh7179+rf/7p169Cn\nTx/Mnz+/1m1y7tw5DBs2DD179sT8+fNRUVEBAPjiiy/wwgsvVJv2ziGg//znP9i9ezc2bNgAPz8/\nTJs2rcb2raysRFxcHPr27Yu+ffsiLi4OlZWV1f7fbNy4Eb1790bfvn3x+eef17ltsrOzMW3aNPTq\n1QuDBg3CZ599BgDYtm0bFi5ciDNnzsDPzw+rV6+uMe+mTZtgb2+PpKQktG3bFgCgVCqxcOHCWkPk\n6NGjCA4ORvfu3fHUU09hzZo1+r6KigpERkbC398fKpUKzz//PPLy8vTba8CAAfDz80NgYCB27dql\nn2/79u0YOnQoevbsicmTJyMjIwPA7cNyS5cuRe/evdG9e3cMHz4cv/32W53bgSQQ9EDo37+/OH78\neI32a9euiU6dOom5c+eKkpISUVZWJrKyskSvXr3E0aNHhVarFd99953o1auXyM/PF0IIMXr0aLF0\n6VJRUVEhfvzxR9GtWzfx+uuvCyGE+OGHH0RAQECd6/7www/FqFGjhFqtFhUVFeKtt94Ss2fPrlbL\nggULRFlZmbhw4YLo0qWL+P3334UQQnzwwQciKChIXL58Weh0OnHhwgWh0WjEL7/8Ivr06SO0Wq0Q\nQoj8/HzRtWtXkZubW+u26NSpk7h69Wq1tsrKSuHj4yOOHj0qhBBi/Pjx4rPPPhNCCDF79myxdu1a\nodVqRXl5uTh16lSdy/rhhx+Ej4+PSExMFBUVFaKsrKzGNunfv7945plnRGZmpigoKBBjxowRK1as\nEEII8fnnn4uxY8fWWW9UVJR+2tq278qVK8WoUaNEXl6eyM/PF2PGjBHvvvtutdpWrlwpKisrxdGj\nR0XXrl1FYWFhrdspNDRUxMTEiPLycnH+/Hnh7+8vvv/++zrrvNuoUaPEqlWr6uz/6/v64YcfxMWL\nF4VWqxUXLlwQvXv3Fl999ZUQQogtW7aIV199VZSWlopbt26Jc+fOiaKiIlFSUiL8/PzE5cuXhRBC\nZGdni99++00IIcRXX30lBg4cKH7//XdRVVUlkpOTxZgxY4QQQnzzzTfi2WefFTdu3BA6nU78/vvv\nIjs7u95aqX4ckTxAZs6cCZVKBZVKhRkzZlTrCw8Ph52dHVq0aIHU1FQ8+eSTeOqppyCXy9GnTx/4\n+vri2LFjyMzMxLlz5xAREQGFQoGePXsiMDBQcg1bt27F7Nmz4enpCYVCgbCwMBw4cAC3bt3STxMW\nFoYWLVrA29sb3t7euHjxIoDbe8IRERF49NFHIZPJ4O3tjVatWqFr165wdHTEiRMnAAB79+5Fr169\n8NBDD0muy8bGBq1atcKNGzdq9FlbWyMzMxM5OTmwtbWFSqWqd1lyuRyvvfYaFAoFWrRoUes048aN\ng1KphIuLC6ZPn95o5zl2796NmTNnws3NDa6urpg5c2a1vXRra2vMnDkTNjY2eOqpp2BnZ4crV67U\nWI5arcZPP/2EyMhI2NrawsfHB6NGjUJqaqqkOgoLC9G6dWvJdfv7+6Nz586Qy+Xw9vbGM888gx9/\n/FFfc2FhIdLT02FlZQVfX184ODgAuL2tL126hPLycri7u+Oxxx4DcPtzNnXqVHTs2BHW1taYNm0a\nLly4gIyMDFhbW6OkpAR//PEHhBDo2LEj3N3dJddKNTW/g7dUp+Tk5DrPkXh6eur/zszMxP79+3Hk\nyBF9261bt+Dv74+cnBw4OTlVOzbu5eUFtVotqYbMzEzMnDkTcvn/9mHkcjny8/P1r+8OgJYtW6K0\ntBQAkJWVhYcffrjW5T777LPYtWsX+vTpg127dmHChAmS6rmjqqoKGo0Gzs7ONfrmzp2LVatWISQk\nBM7Ozpg4cWK9Vyq1atUKtra29a5PqVTq//by8kJOTk6D6q1LTk4OvLy86ly2i4tLtXM2d2/fvy7H\n2dlZ/4V9Z1lpaWmS6nBxcUFubq7kun/55RcsX74cly5dQlVVFSorKzFkyBAAtw8dZmVlYc6cObh5\n8yZGjBiB2bNnw87ODu+++y42btyIBQsWoHv37oiKikLHjh2RmZmJpUuXIiEhQb8OIQSys7PRu3dv\njBs3DrGxscjIyMDTTz+NqKioau+VGoZBQgAAmUym/1upVGLkyJFYsmRJjekyMjJw8+ZNlJaW6sMk\nMzNTP3/Lli1RXl6un16r1UKj0ehfe3p6YunSpejRo0eNZV+/fr3eGj09PfHnn3+iU6dONfpGjBiB\noKAgXLx4EZcvX8bAgQPv8Y6rO3ToEKysrNC1a9cafa1bt9Zvi9OnT2PixIno2bNnnVdq3b0t63J3\n8GZmZur3iP+6/f76ZXyvZbu7uyMzM1O/Z65Wq+9rb9vd3R03btxAcXGx/gtWrVZLvsqvd+/e+Oqr\nrxAWFlZtp6Eur7/+OsaPH4/169fD1tYWcXFx+vNVNjY2CAsLQ1hYGK5fv46pU6eiQ4cOGDVqFAIC\nAhAQEIDy8nKsXLkSb731Fj799FMolUpMmzYNI0aMqHV9EyZMwIQJE5Cfn49Zs2Zh/fr1+nNf1HA8\ntEU1jBgxAkeOHMG3334LrVaLiooKnDx5EllZWWjTpg18fX2xZs0aVFZW4vTp09VGLh06dEBFRQWO\nHj2KqqoqvPfee/qTvQDwwgsvYOXKlfoTnxqNBl9//bWkukaNGoVVq1bh6tWrEELg4sWL+i8bT09P\n/P3vf8fcuXPx9NNP13lI6a8KCwuxa9cuxMbG4pVXXkGrVq1qTLNv3z5kZWUBAJydnSGTyfRfjg89\n9BCuXbsmaV13+/TTT5GVlYXCwkKkpKToLz329vbGpUuXcOHCBVRUVFQ76QwAbm5u9QbuM888g/fe\new8ajQYajQbJyckYPnx4g+tTKpXw8/PDihUrUFFRgYsXL2L79u11fjH/1cSJE1FSUoKoqCj9/+vs\n7GzEx8frD1XeraSkBM7OzrC1tcXZs2exZ88efd8PP/yAX3/9FVqtFg4ODrC2toZcLkdeXh6+/vpr\nlJaWQqFQwM7OTv//ZezYsVi3bh0uXboEACgqKsK+ffsAAGfPnsUvv/yCqqoqtGzZEgqFQlLYUd04\nIqEalEol1q5di6SkJLz++uuQy+Xo2rUrFi1aBAB45513EBUVBX9/f3Tr1g3BwcG4efMmAMDR0REx\nMTFYuHAhtFotpkyZUu2w2YQJEyCEwKRJk5CTkwM3NzcMGzZM0ghi4sSJqKysxKRJk1BQUIBHH30U\nycnJ+v7g4GC88cYbWLBgwT2XNXLkSMhkMtjY2KBz586YP39+nV+4586dw9KlS1FcXAw3NzcsWLAA\n7dq1A3D7fM68efNQXl6O2NhYuLm53XPdABAUFKTfBgMGDMD06dMB3A7imTNn4uWXX0aLFi0wZ84c\n/Oc//9HPFxISgoiICKhUKvTq1Qtr166tttwZM2agpKRE/4U/ZMiQGufDpFqxYgViYmIQEBAAJycn\nhIeH3/Py8TtcXFywZcsWrFy5EqNHj0ZpaSk8PDwQFBRU60guJiYGCQkJiI2NRa9evTB06FD9Zyov\nLw8xMTHIzs6GnZ0dhg0bhpEjR0Kj0eDDDz9EVFQUZDIZfHx89J/RQYMGoaSkBHPmzEFGRgYcHR3x\nz3/+E0OHDkVJSQmWLl2K69evQ6FQoG/fvpg8efJ9bSO6TSYEH2xFhlmzZg3S09OxfPlyk9Zx6tQp\nzJ07F0eOHJF0eImIGgfHc9QsVFVV4eOPP0ZISAhDhKiJMUjI4l2+fBk9e/ZEbm4uXn75ZVOXQ/TA\n4aEtIiIyCEckRERkEAYJEREZhEFCREQGafa/IykoKIFOx9NARERSyOUytGpl36B5mn2Q6HSCQUJE\nZEQ8tEVERAZhkBARkUEYJEREZBAGCRERGYRBQkREBmGQEBGRQRgkRERkkGb/O5LatLRTwMqq+WWo\nVqtDWWnlvSckImpED2SQWFnJMTsh1dRlNLp3o0aaugQiegA1v91yIiJqUk02IpkxYwauX78OuVwO\nOzs7vPXWW/Dx8UFgYCAUCgVsbW0BAJGRkQgICAAAnDlzBtHR0aioqECbNm2QlJQk+ZnYRETUNJos\nSBISEuDo6AgA+Prrr/Hmm29ix44dAIDVq1ejU6dO1abX6XSYO3cu4uPjoVKpsHbtWixfvhzx8fFN\nVTIREUnQZIe27oQIABQXF9/zudppaWmwtbWFSqUCAIwdOxb79+83ao1ERNRwTXqyfcGCBTh+/DiE\nEFi/fr2+PTIyEkII9OjRA3PmzIGTkxPUajW8vLz007i6ukKn06GwsBAuLi5NWTYREdWjSYMkLi4O\nALBz504kJibigw8+wObNm6FUKlFZWYm4uDjExsZi+fLljbZONzeHGm3lFVWQy5vfdQYyGdC6teO9\nJyQiakQmufw3ODgY0dHRKCgogFKpBAAoFAqEhoZi+vTpAAClUonMzEz9PBqNBnK5vMGjkfz84hrP\nI3FwbAGdTmfguzA/QgC5uUWmLoOILJhcLqt1B7zeeYxUSzUlJSVQq9X614cPH4azszNsbW1RVHT7\ni08Igb1798LHxwcA4Ovri/Lycpw+fRoAsHXrVgwZMqQpyiUiogZokhFJWVkZIiIiUFZWBrlcDmdn\nZ6SkpCA/Px/h4eHQarXQ6XTo2LEjYmJiAAByuRyJiYmIiYmpdvkvERGZF5kQolk/h7auQ1vN9Zft\nxUXlpi6DiCyY2R7aIiKi5otBQkREBmGQEBGRQRgkRERkEAYJEREZhEFCREQGYZAQEZFBGCRERGQQ\nBgkRERmEQUJERAZhkBARkUEYJEREZBAGCRERGYRBQkREBmGQEBGRQRgkRERkEAYJEREZhEFCREQG\nYZAQEZFB7itIrl27huvXrzd2LUREZIEkBcmcOXPw008/AQA+//xzPPPMMwgKCsK2bdskr2jGjBkY\nMWIEgoODERoaigsXLgAArly5gjFjxmDw4MEYM2YMrl69qp+nvj4iIjIPkoLkxIkT8PX1BQB8+OGH\n2LRpE7Zt24YPPvhA8ooSEhKwa9cu7Ny5E5MmTcKbb74JAIiJiUFoaCgOHDiA0NBQREdH6+epr4+I\niMyDpCCpqqqCQqFAdnY2CgsL0aNHDzz22GPIy8uTvCJHR0f938XFxZDJZMjPz8f58+cRFBQEAAgK\nCsL58+eh0Wjq7SMiIvNhLWUiHx8fvP/++8jIyEC/fv0AANnZ2XBwcGjQyhYsWIDjx49DCIH169dD\nrVbDw8MDVlZWAAArKyu4u7tDrVZDCFFnn6urq+R1urnVrLG8ogpyefO7zkAmA1q3drz3hEREjUhS\nkMTFxWHVqlWwtrbGG2+8AQD4+eefMXz48AatLC4uDgCwc+dOJCYmIiIiooHlNlx+fjF0OlGtzcGx\nBXQ6ndHX3dSEAHJzi0xdBhFZMLlcVusOeH0kBcnDDz+Md955p1rbkCFDMGTIkAat7I7g4GBER0fD\n09MT2dnZ0Gq1sLKyglarRU5ODpRKJYQQdfYREZH5kHR8RwiBzz77DC+99JJ+FHLq1Cns3btX0kpK\nSkqgVqv1rw8fPgxnZ2e4ubnBx8cHe/bsAQDs2bMHPj4+cHV1rbePiIjMh6QRyapVq/D999/jpZde\nQkxMDADA09MT8fHxGDZs2D3nLysrQ0REBMrKyiCXy+Hs7IyUlBTIZDIsWrQI8+bNw9q1a+Hk5ISE\nhAT9fPX1ERGReZAJIcS9JnrqqaewY8cOuLq6omfPnjh16hSEEOjVqxdOnTrVFHXet7rOkcxOSDVR\nRcbzbtRIFBeVm7oMIrJg93OORNKhLa1WC3t7ewCATCYDcPtwlZ2dXQNLJCKi5kZSkDz11FOIj49H\nZWUlgNvnTFatWoX+/fsbtTgiIjJ/koJk/vz5yM3NRY8ePVBUVAQ/Pz9kZmYiMjLS2PUREZGZk3Sy\n3cHBAcnJycjPz0dGRgaUSiVat25t7NqIiMgCSAqS7777Dm3atEGHDh3g5uYGAPjjjz+gVqvRp08f\noxZIRETmTdKhrdjYWP3J9jvs7e0RGxtrlKKIiMhySAqS/Px8uLu7V2tzd3dHbm6uUYoiIiLLISlI\n2rVrhxMnTlRrO3nyJNq2bWuUooiIyHJIOkcSFhaG8PBwhISEoF27drh27Rq++OILLF261Nj1ERGR\nmZM0Ihk4cCA2btyI0tJSHDt2DKWlpVi/fj0GDhxo7PqIiMjMSRqRAEDXrl3RtWtXY9ZCREQWSFKQ\nVFZWYseOHbhw4QJKS0ur9SUmJhqlMCIisgySgmTevHm4ePEi+vfvj4ceesjYNRERkQWRFCTffvst\nDh06BCcnJ2PXQ0REFkbSyXalUqm/YSMREdHdJI1IgoODMWPGDEyYMEF/i5Q7evfubZTCiIjIMkgK\nkn//+98AgBUrVlRrl8lkOHToUONXRUREFkNSkBw+fNjYdRARkYWSdI6EiIioLpJGJMXFxVizZg1O\nnTqFgoIC3P2Y96NHj95z/oKCArzxxhv4888/oVAo8MgjjyA2Nhaurq7o3LkzOnXqBLn8dqYlJiai\nc+fOAG6PhBITE6HVatGlSxfEx8ejZcuW9/E2iYjIWCSNSBYtWoTz589jxowZKCwsxMKFC6FUKvHy\nyy9LWolMJsOUKVNw4MAB7N69G+3atcPy5cv1/Vu3bkVqaipSU1P1IVJSUoK33noLKSkp+Oqrr2Bv\nb48NGzY0/B0SEZFRSQqS48ePY/Xq1Rg4cCCsrKwwcOBArFy5EqmpqZJW4uLiAn9/f/3rbt26ITMz\ns955vvnmG/j6+qJ9+/YAgLFjx2Lfvn2S1kdERE1H0qEtnU4HR0dHAICdnR2KiorQunVrpKenN3iF\nOp0OW7ZsQWBgoL7txRdfhFarxZNPPonw8HAoFAqo1Wp4eXnpp/Hy8oJarW7w+oiIyLgkBYm3tzdO\nnTqF3r17Q6VSYdGiRbC3t9ePFhpi8eLFsLOzw/jx4wHcPseiVCpRXFyMuXPnIjk5GbNnz27wcuvi\n5uZQo628okp/TqY5kcmA1q0dTV0GET1gJAXJkiVL9CfYFyxYgBUrVuDmzZsNvmFjQkIC0tPTkZKS\nov8iVyqVAAAHBweMGjUKmzZt0refPHlSP29mZqZ+2obIzy+GTieqtTk4toBOp2vwssydEEBubpGp\nyyAiCyaXy2rdAa93nntNoNVq8cUXX8DT0xMA4Obmhri4OKxcuRJ/+9vfJK9oxYoVSEtLQ3JyMhQK\nBQDgxo0bKC8vBwDcunULBw4cgI+PDwAgICAA586dw9WrVwHcPiE/dOjQBr05IiIyvnuOSKysrPDp\np58iPDz8vldy6dIlvP/++2jfvj3Gjh0LAGjbti2mTJmC6OhoyGQy3Lp1C35+foiIiABwe4QSGxuL\nV199FTqdDj4+PliwYMF910BERMYhE3f/KKQO8fHxePjhhzFu3LimqKlR1XVoa3aCtCvOLMm7USNR\nXFRu6jKIyILdz6EtSedIzp49i3//+9/YsGEDPD09IZPJ9H2bN29uWJVERNSsSAqS0aNHY/To0cau\nhYiILJCkIHn22WeNXQcREVkoSUGyffv2OvtCQkIarRgiIrI8koLkr7dCycvLw7Vr1+Dn58cgISJ6\nwEkKkk8++aRG2/bt23H58uVGL4iIiCzLfd8n5LnnnsPnn3/emLUQEZEFknzTxruVlZVh165d+hs5\nEhHRg0tSkDz++OPVfjsCAB4eHoiNjTVKUUREZDkkBcmhQ4eqvW7ZsiVcXV2NUhAREVkWSUFibW2N\nFi1awNnZWd9254aLHh4eRiuOiIjMn6ST7TNmzEBWVla1tqysLISFhRmlKCIishySguTKlSv6Z6nf\n0blzZ/zxxx9GKYqIiCyHpCBxc3Or8Vjd9PR0uLi4GKUoIiKyHJKC5Pnnn0d4eDiOHDmC33//HYcP\nH8Zrr72GUaNGGbs+IiIyc5JOtk+dOhXW1tZISEhAVlYWlEolQkJCMHHiRGPXR0REZk5SkMjlckyZ\nMgVTpkwxdj1ERGRhJB3aWrduHc6ePVut7ezZs/jggw+MUhQREVkOSUHy8ccf429/+1u1to4dO+Kj\njz4ySlFERGQ5JAVJVVUVrK2rHwWzsbFBZWWlUYoiIiLLISlIunTpgk8//bRa29atW/H4449LWklB\nQQFeeeUVDB48GMOHD0dYWBg0Gg0A4MyZMxgxYgQGDx6MSZMmIT8/Xz9ffX1ERGQeZEIIca+JLl26\nhIkTJ8Ld3R3t2rXDtWvXkJubi02bNtU45FWbwsJC/Prrr/D39wcAJCQk4MaNG1iyZAkGDx6M+Ph4\nqFQqrF27FteuXUN8fDx0Ol2dfQ2Rn18Mna76W3RwbIHZCal1zGG53o0aieKiclOXQUQWTC6Xwc3N\noWHzSJnosccew4EDBzB58mT8/e9/x+TJk7F//35JIQIALi4u+hABgG7duiEzMxNpaWmwtbWFSqUC\nAIwdOxb79+8HgHr7iIjIfEi6/BcAcnNz4eXlhS5duqB9+/b3vUKdToctW7YgMDAQarUaXl5e+j5X\nV1fodDoUFhbW28df1BMRmY97BsnBgwexbNkyZGZmAgBkMhmUSiXeeOMNDBkypMErXLx4Mezs7DB+\n/Hh89dVXDa+4gWobopVXVEEuv++HQ5otmQxo3ZoPG7tf5VUVpi7BaFrY2Jq6BGrG6g2So0ePYv78\n+Zg2bRqGDh0Kd3d35OTkYO/evVi4cCFsbW3Rv39/yStLSEhAeno6UlJSIJfLoVQq9QEFABqNBnK5\nHC4uLvX2NURd50j++tTH5kAIIDe3yNRlWCwHZxtEbV9i6jIaXULIQn4uSLJGP0eydu1axMbG4pVX\nXkHbtm2hUCjQtm1bTJ06FYsWLcLatWslr2jFihVIS0tDcnIyFAoFAMDX1xfl5eU4ffo0gNtXgt0Z\n5dTXR0RE5qPeEcmlS5cwaNCgWvuefvppvPXWW5JWcunSJbz//vto3749xo4dCwBo27YtkpOTkZiY\niJiYGFRUVKBNmzZISkoCcPu2LHX1ERGR+ag3SBQKBYqLi2t9rO7Nmzf1I4t7eeyxx/Drr7/W2te9\ne3fs3r27wX1ERGQe6j20FRAQgHfeeafWvhUrVqBv375GKYqIiCxHvSOSuXPn4oUXXsDw4cMxePBg\ntG7dGrm5uTh48CCKi4tr/NqdiIgePPUGiYeHB3bs2IFNmzbh22+/RUFBAVq1aoXAwEC8/PLL/D0H\nERHd+3ckzs7OmDVrFmbNmtUU9RARkYVpfr/KIyKiJsUgISIigzBIiIjIIHUGyejRo/V//+tf/2qS\nYoiIyPLUGSRXr15FRcXtm9ht3LixyQoiIiLLUudVWwMGDMDgwYPRpk0bVFRUYNy4cbVOt3nzZqMV\nR0RE5q/OIImPj8fp06eRkZGBc+fOISQkpCnrIiIiC1Hv70hUKhVUKhWqqqrw7LPPNlVNRERkQSQ9\nITEkJAQnT57Ezp07kZOTA3d3d4wcORJPPPGEsesjIiIzJ+ny323btmHWrFlo3bo1Bg0aBHd3d7z+\n+uv47LPPjF0fERGZOUkjkvXr12PTpk3w9vbWtw0dOhSvvfZatcuEiYjowSNpRFJYWIiOHTtWa3v0\n0Udx48YNoxRFRESWQ1KQdO/eHcuWLUNZWRkAoLS0FImJifDz8zNqcUREZP4kHdp6++23MXv2bKhU\nKjg7O+PGjRvw8/Or86FXRET04JAUJO7u7ti8eTOysrL0V215enoauzYiIrIAkoLkDk9PTwYIERFV\n06AgMURCQgIOHDiAjIwM7N69G506dQIABAYGQqFQwNbWFgAQGRmJgIAAAMCZM2cQHR2NiooKtGnT\nBklJSXBzc2uqkomISIImu438gAEDsHnzZrRp06ZG3+rVq5GamorU1FR9iOh0OsydOxfR0dE4cOAA\nVCoVli9f3lTlEhGRRPcMEp1OhxMnTqCystKgFalUKiiVSsnTp6WlwdbWFiqVCgAwduxY7N+/36Aa\niIio8d3z0JZcLseMGTPw888/G62IyMhICCHQo0cPzJkzB05OTlCr1fDy8tJP4+rqCp1Oh8LCQri4\nuEhetpubQ4228ooqyOXN75nRpC/8AAARpUlEQVReMhnQurWjqcuwWOVVFZDLZaYuo9HJZDJ+Lsio\nJJ0j6dmzJ86cOYNu3bo1egGbN2+GUqlEZWUl4uLiEBsb26iHsPLzi6HTiWptDo4toNPpGm0d5kII\nIDe3yNRlWCwHZ5san5XmQAjBzwVJJpfLat0Br4+kIPHy8sIrr7yCAQMGwNPTEzLZ//baIiIiGlbl\nX9w53KVQKBAaGorp06fr2zMzM/XTaTQayOXyBo1GiIjI+CQFSUVFBQYOHAgAyM7ObrSVl5aWQqvV\nwtHREUII7N27Fz4+PgAAX19flJeX4/Tp01CpVNi6dSuGDBnSaOsmIqLGISlI4uPjDV7RkiVLcPDg\nQeTl5WHixIlwcXFBSkoKwsPDodVqodPp0LFjR8TExAC4fW4mMTERMTEx1S7/JSIi8yL5dySXL1/G\n/v37kZ+fj+joaPzxxx+orKysdkfg+ixcuBALFy6s0b5z58465+nevTt2794ttUQiIjIBSZcu7du3\nD+PGjUN2drb+i7+kpATLli0zanFERGT+JI1IVq9ejQ8//BDe3t7Yt28fAMDb2xsXL140anFERGT+\nJI1INBoNOnfuDAD6K7ZkMlm1q7eIiOjBJClIunTpgtTU1GptX375Jbp27WqUooiIyHJIOrS1YMEC\nTJ48Gdu3b0dpaSkmT56MK1euYOPGjcauj4iIzJykIOnYsSP27duHI0eOoF+/flAqlejXrx/s7e2N\nXR8REZk5yZf/tmzZEj169EDbtm3h4eHBECEiIgASgyQzMxORkZH45Zdf4OTkhJs3b+If//gHkpKS\nar0tPBERPTgknWyPiopCly5dcOrUKZw4cQI//vgjfH19MW/ePGPXR0REZk7SiOS///0vNm7cCBsb\nGwCAvb09IiMj4e/vb9TiiIjI/EkakXTr1g1nz56t1paWlgY/Pz+jFEVERJajzhHJqlWr9H+3a9cO\nU6dORb9+/eDp6YmsrCwcO3YMQUFBTVIkERGZrzqDJCsrq9rrp59+GsDtX7krFAoMGjQIFRUVxq2O\niIjMXp1B0hi3jiciouZP8u9IysrKkJ6ejtLS0mrt3bt3b/SiiIjIckgKkp07dyI2NhY2NjZo0aKF\nvl0mk+Ho0aPGqo2IiCyApCBJSkrCmjVr0KdPH2PXQ0REFkbS5b82Njbo1auXsWshIiILJClIIiIi\nsGzZMmg0GmPXQ0REFkZSkLRv3x6HDx9Gnz594OPjAx8fH3h7e8PHx0fSShISEhAYGIjOnTvjt99+\n07dfuXIFY8aMweDBgzFmzBhcvXpVUh8REZkPSedI3njjDYwcORLDhg2rdrJdqgEDBmDChAkYN25c\ntfaYmBiEhoZi5MiRSE1NRXR0ND7++ON79hERkfmQNCIpLCxEREQEOnXqhIcffrjaf1KoVCoolcpq\nbfn5+Th//rz+1/FBQUE4f/48NBpNvX1ERGReJAXJc889V+NRu4ZSq9Xw8PCAlZUVAMDKygru7u5Q\nq9X19hERkXmRdGjr7Nmz2Lx5M9577z089NBD1fo2b95slMIai5ubQ4228ooqyOWSMtSiyGRA69aO\npi7DYpVXVUAul5m6jEYnk8n4uSCjkhQko0ePxujRoxt1xUqlEtnZ2dBqtbCysoJWq0VOTg6USiWE\nEHX2NVR+fjF0OlGtzcGxBXQ6XWO9FbMhBJCbW2TqMiyWg7NNjc9KcyCE4OeCJJPLZbXugNdHUpA8\n++yz91VQfdzc3ODj44M9e/Zg5MiR2LNnD3x8fODq6goA9fYREZH5kBQk27dvr7MvJCTknvMvWbIE\nBw8eRF5eHiZOnAgXFxd8+eWXWLRoEebNm4e1a9fCyckJCQkJ+nnq6yMiIvMhKUj+eqI9Ly8P165d\ng5+fn6QgWbhwIRYuXFijvWPHjti2bVut89TXR0RE5kNSkHzyySc12rZv347Lly83ekFERGRZ7vvS\npeeeew6ff/55Y9ZCREQWSNKI5K9XOJWVlWHXrl1wdOQlhUREDzpJQfL4449DJqt+fb2HhwcWL15s\nlKKIiMhySAqSQ4cOVXvdsmVLXopLREQAJAZJmzZtjF0HERFZqHqD5MUXX6xxSOtuMpkMH330UaMX\nRURElqPeIBkxYkSt7dnZ2fjkk09QXl5ulKKIiMhy1Bsko0aNqva6oKAA69atw2effYZhw4Zh5syZ\nRi2OiIjMn6RzJMXFxVi/fj02b96Mfv36YceOHZKfRUJERM1bvUFSXl6Ojz76CBs3boS/vz8+/fRT\nPPbYY01VGzUBZ3sryK2a3y31dVodbpRoTV0G0QOh3iAJDAyETqfDlClT4Ovri7y8POTl5VWbpnfv\n3kYtkIxLbiVH2ro3TV1Go/OduhQAg4SoKdQbJHeez75ly5Za+2UyWY3fmBAR0YOl3iA5fPhwU9VB\nREQWqvkdHCcioibFICEiIoMwSIiIyCAMEiIiMoikHyQSET2o7O0VsJI3v31urU6HkpLKRlkWg4SI\nqB5WcjneW7Hf1GU0uulzhjTasswiSAIDA6FQKGBrawsAiIyMREBAAM6cOYPo6GhUVFSgTZs2SEpK\ngpubm4mrJSKiu5lFkADA6tWr0alTJ/1rnU6HuXPnIj4+HiqVCmvXrsXy5csRHx9vwiqJiOivzPbA\nX1paGmxtbaFSqQAAY8eOxf79zW94SURk6cxmRBIZGQkhBHr06IE5c+ZArVbDy8tL3+/q6gqdTofC\nwkK4uLiYsFIiIrqbWQTJ5s2boVQqUVlZibi4OMTGxmLQoEGNsmw3N4cabeUVVZA3w6swZDKgdWvH\nBs2jrSyHXF73UzAt1f1si/Kqima6LWQN3hb0P5UVt/i5uAezCBKlUgkAUCgUCA0NxfTp0zFhwgRk\nZmbqp9FoNJDL5Q0ejeTnF0OnE9XaHBxbQKfTGV64mRECyM0tatA8rZxsamyf5uB+toWDc3PdFqLB\n24L+x8mxxQP1uZDLZbXugNfH5LvlpaWlKCq6/WaEENi7dy98fHzg6+uL8vJynD59GgCwdetWDBnS\neJerERFR4zD5iCQ/Px/h4eHQarXQ6XTo2LEjYmJiIJfLkZiYiJiYmGqX/xIRkXkxeZC0a9cOO3fu\nrLWve/fu2L17dxNXREREDWHyQ1tERGTZTD4iISLz49jSClZWzW8/U6vVoaiMj2BubAwSIqrBykqO\no3OjTF1Go+uXlACAQdLYmt8uBxERNSkGCRERGYRBQkREBmGQEBGRQRgkRERkEAYJEREZhEFCREQG\nYZAQEZFBGCRERGQQBgkRERmEQUJERAZhkBARkUEYJEREZBAGCRERGYRBQkREBmGQEBGRQRgkRERk\nELMPkitXrmDMmDEYPHgwxowZg6tXr5q6JCIiuovZB0lMTAxCQ0Nx4MABhIaGIjo62tQlERHRXcz6\nme35+fk4f/48Nm3aBAAICgrC4sWLodFo4OrqKmkZcrmsRptMBrg52zVqreZAJqv9/d5jLigcWxml\nHtOSNXhbyCCDq72LkeoxHdl9bAvIZGgh8d+YRZHdz7YAnJxbGqceU6rj+6Lh3yGATAghGqMmY0hL\nS0NUVBS+/PJLfduwYcOQlJSELl26mLAyIiK6w+wPbRERkXkz6yBRKpXIzs6GVqsFAGi1WuTk5ECp\nVJq4MiIiusOsg8TNzQ0+Pj7Ys2cPAGDPnj3w8fGRfH6EiIiMz6zPkQDA5cuXMW/ePNy8eRNOTk5I\nSEjAo48+auqyiIjo/zP7ICEiIvNm1oe2iIjI/DFIiIjIIAwSIiIyCIOEiIgMwiAxEt5s8n8SEhIQ\nGBiIzp0747fffjN1OSZTUFCAV155BYMHD8bw4cMRFhYGjUZj6rJMZsaMGRgxYgSCg4MRGhqKCxcu\nmLokk/vXv/5lkf9OGCRGwptN/s+AAQOwefNmtGnTxtSlmJRMJsOUKVNw4MAB7N69G+3atcPy5ctN\nXZbJJCQkYNeuXdi5cycmTZqEN99809QlmdR///tfnDlzxiL/nTBIjODOzSaDgoIA3L7Z5Pnz5x/Y\nvU+VSsW7EQBwcXGBv7+//nW3bt2QmZlpwopMy9HRUf93cXExZLKG3yywuaisrERsbCwWLVpk6lLu\ni1nf/ddSqdVqeHh4wMrKCgBgZWUFd3d3qNVq/iqfAAA6nQ5btmxBYGCgqUsxqQULFuD48eMQQmD9\n+vWmLsdkVq1ahREjRqBt27amLuW+cERCZAKLFy+GnZ0dxo8fb+pSTCouLg5Hjx7F7NmzkZiYaOpy\nTOLnn39GWloaQkNDTV3KfWOQGAFvNkn1SUhIQHp6OlauXAm5nP8EASA4OBgnT55EQUGBqUtpcqdO\nncLly5cxYMAABAYGIisrC5MnT8Z3331n6tIk46fYCHizSarLihUrkJaWhuTkZCgUClOXYzIlJSVQ\nq9X614cPH4azszNcXJrfg8XuZerUqfjuu+9w+PBhHD58GJ6entiwYQP69u1r6tIk4722jIQ3m/yf\nJUuW4ODBg8jLy0OrVq3g4uJS7WFlD4pLly4hKCgI7du3R4sWLQAAbdu2RXJysokra3p5eXmYMWMG\nysrKIJfL4ezsjKioKD6wDkBgYCBSUlLQqVMnU5ciGYOEiIgMwkNbRERkEAYJEREZhEFCREQGYZAQ\nEZFBGCRERGQQBgmRAdasWYPIyEhTl0FkUrzXFpEEu3fvxqZNm3DlyhXY29vD29sb06ZNM3VZRGaB\nQUJ0D5s2bcK6devw9ttvo2/fvrCxscG3336LQ4cOwc7OztTlEZkcD20R1aOoqAirV69GdHQ0nn76\nadjZ2cHGxgaBgYGIioqqMf1rr72GPn36oEePHhg3bhwuXbqk7zt27BiGDRsGPz8/BAQEYMOGDQAA\njUaDV199FSqVCr169UJoaCh0Oh0AIDs7G+Hh4XjiiScQGBiIjz/+WL+8s2fP4rnnnkP37t3xz3/+\nE/Hx8UbeGkS144iEqB4///wzKioqMGjQIEnTP/nkk1i6dCkUCgWSkpIQGRmJ1NRUALdvmb5y5Uqo\nVCrcuHED169fB3B7xOPh4YETJ04AAH755RfIZDLodDpMnz4dgYGBeOedd5CdnY2XX34ZHTp0QEBA\nAOLi4jBhwgQEBwejpKSkWmgRNSWOSIjqUVhYiFatWsHaWto+V0hICBwcHKBQKBAeHo6LFy+iqKgI\nAGBtbY3ff/8dxcXFcHZ21t9XytraGrm5ucjMzISNjQ1UKhVkMhnOnTsHjUaDsLAwKBQKtGvXDqNH\nj8bevXv18/3555/QaDSwt7dHt27djLMRiO6BQUJUDxcXFxQUFODWrVv3nFar1WL58uUYOHAgunfv\nrn9o1Z1bo69evRrHjh1D//79MX78ePz8888AgMmTJ+ORRx7BpEmTMGDAAKxbtw4AkJGRgZycHKhU\nKv1/KSkpyMvLA3D7WR5Xr17F0KFD8fzzz+PIkSPG2ARE9yaIqE43b94U//jHP8S+fftq7V+9erV4\n/fXXhRBC7NixQwwZMkT8+eefQqfTiRs3bohOnTqJq1evVpunsrJSbNq0STz55JM1lvfrr7+KJ554\nQnz//ffip59+EoMGDbpnjVqtVuzbt0/4+vqKkpKS+3iXRIbhiISoHo6OjnjttdcQGxuLr7/+GmVl\nZaiqqsKxY8dqPNGvpKQECoUCrVq1QllZGVasWKHvq6ysxK5du1BUVAQbGxvY29vrH2p15MgRpKen\nQwgBR0dHWFlZQSaToWvXrrC3t8e6detQXl4OrVaL3377DWfPngUApKamQqPRQC6Xw8nJCQD4oCwy\nCZ5sJ7qHSZMm4aGHHsLatWsRGRkJe3t7dOnSBdOmTcPx48f10wUHB+O7775DQEAAXFxcEBERgS1b\ntuj7U1NTsXjxYmi1WnTo0AFJSUkAgPT0dCxevBgajQZOTk544YUX8MQTTwAAUlJSkJCQgAEDBqCy\nshIdOnTArFmzAADffvstli1bhvLycnh5eeHdd9/VP+eEqCnxeSRERGQQjoOJiMggDBIiIjIIg4SI\niAzCICEiIoMwSIiIyCAMEiIiMgiDhIiIDMIgISIigzBIiIjIIP8PAlE1WXkr+HUAAAAASUVORK5C\nYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "Class_Frequency_Count = val_dataframe['diagnosis'].value_counts()\n",
    "print(str(Class_Frequency_Count) + \"\\n\")\n",
    "sns.set(style=\"darkgrid\")\n",
    "sns.barplot(Class_Frequency_Count.index,Class_Frequency_Count.values,alpha=0.9)\n",
    "plt.title(\"Frequency Distribution of Classes\")\n",
    "plt.ylabel(\"Number of Occurances\", fontsize=12)\n",
    "#plt.xticks(rotation=90)\n",
    "plt.xlabel(\"Classes\", fontsize=12)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "42MdRPIiuFwa"
   },
   "outputs": [],
   "source": [
    "import shutil as cf\n",
    "for keyword in keywords:\n",
    "  image_list = []\n",
    "  class_list = []\n",
    "  if(not os.path.exists(keyword)):\n",
    "    os.mkdir(keyword)\n",
    "  if keyword == train_keyword:\n",
    "    image_list = train_images\n",
    "    class_list = train_labels\n",
    "  elif keyword == validation_keyword: \n",
    "    image_list = val_images\n",
    "    class_list = val_labels\n",
    "  for image_info in image_list:\n",
    "    cl = class_list[image_list.index(image_info)]\n",
    "    cf.copy('Train_Images/'+str(image_info)+'.png', keyword+'/'+str(cl))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "H2meYmoF2L5V"
   },
   "outputs": [],
   "source": [
    "!rm -rf Train_Images"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "9WOmXnS8SoSn"
   },
   "source": [
    "# **Image Data Generator**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 68
    },
    "colab_type": "code",
    "id": "MgmDorT-StSa",
    "outputId": "ee4b1fd1-c4ba-430e-9a4e-e09b91782686"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.14.0\n",
      "Found 3127 images belonging to 5 classes.\n",
      "Found 733 images belonging to 5 classes.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import keras\n",
    "import numpy as np\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Dropout\n",
    "from keras.layers import Flatten\n",
    "from keras.layers import Activation\n",
    "from keras.layers.convolutional import Conv2D\n",
    "from keras.layers.convolutional import MaxPooling2D\n",
    "from keras.optimizers import SGD\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "print(tf.__version__)\n",
    "\n",
    "train_dir = 'Training_Dir'\n",
    "validation_dir = 'Validation_Dir'\n",
    "\n",
    "train_datagen = ImageDataGenerator(\n",
    "      rescale=1./255,\n",
    "      rotation_range=10,\n",
    "      width_shift_range=0.3,\n",
    "      height_shift_range=0.3,\n",
    "      horizontal_flip=True,\n",
    "      fill_mode='nearest')\n",
    "\n",
    "\n",
    "validation_datagen = ImageDataGenerator(rescale=1./255)\n",
    " \n",
    "# Change the batchsize according to your system RAM\n",
    "train_batchsize = 120\n",
    "val_batchsize = 40\n",
    " \n",
    "train_generator = train_datagen.flow_from_directory(\n",
    "        train_dir,\n",
    "        target_size=(299, 299),\n",
    "        batch_size=train_batchsize,\n",
    "        class_mode='categorical')\n",
    " \n",
    "validation_generator = validation_datagen.flow_from_directory(\n",
    "        validation_dir,\n",
    "        target_size=(299, 299),\n",
    "        batch_size=val_batchsize,\n",
    "        class_mode='categorical',\n",
    "        shuffle=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "by8NP_t0M8rX"
   },
   "source": [
    "# **Fine Tuning Model**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "gLE_ketd-Won"
   },
   "outputs": [],
   "source": [
    "from keras.applications.inception_resnet_v2 import InceptionResNetV2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "9yHILp1ReCeR",
    "outputId": "ead10d56-4dd3-4552-e149-8db05356d2df"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_6 (InputLayer)            (None, 299, 299, 3)  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1016 (Conv2D)            (None, 149, 149, 32) 864         input_6[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1016 (Batch (None, 149, 149, 32) 96          conv2d_1016[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1016 (Activation)    (None, 149, 149, 32) 0           batch_normalization_1016[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1017 (Conv2D)            (None, 147, 147, 32) 9216        activation_1016[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1017 (Batch (None, 147, 147, 32) 96          conv2d_1017[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1017 (Activation)    (None, 147, 147, 32) 0           batch_normalization_1017[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1018 (Conv2D)            (None, 147, 147, 64) 18432       activation_1017[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1018 (Batch (None, 147, 147, 64) 192         conv2d_1018[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1018 (Activation)    (None, 147, 147, 64) 0           batch_normalization_1018[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_21 (MaxPooling2D) (None, 73, 73, 64)   0           activation_1018[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1019 (Conv2D)            (None, 73, 73, 80)   5120        max_pooling2d_21[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1019 (Batch (None, 73, 73, 80)   240         conv2d_1019[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1019 (Activation)    (None, 73, 73, 80)   0           batch_normalization_1019[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1020 (Conv2D)            (None, 71, 71, 192)  138240      activation_1019[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1020 (Batch (None, 71, 71, 192)  576         conv2d_1020[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1020 (Activation)    (None, 71, 71, 192)  0           batch_normalization_1020[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_22 (MaxPooling2D) (None, 35, 35, 192)  0           activation_1020[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1024 (Conv2D)            (None, 35, 35, 64)   12288       max_pooling2d_22[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1024 (Batch (None, 35, 35, 64)   192         conv2d_1024[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1024 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1024[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1022 (Conv2D)            (None, 35, 35, 48)   9216        max_pooling2d_22[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1025 (Conv2D)            (None, 35, 35, 96)   55296       activation_1024[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1022 (Batch (None, 35, 35, 48)   144         conv2d_1022[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1025 (Batch (None, 35, 35, 96)   288         conv2d_1025[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1022 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1022[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1025 (Activation)    (None, 35, 35, 96)   0           batch_normalization_1025[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "average_pooling2d_6 (AveragePoo (None, 35, 35, 192)  0           max_pooling2d_22[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1021 (Conv2D)            (None, 35, 35, 96)   18432       max_pooling2d_22[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1023 (Conv2D)            (None, 35, 35, 64)   76800       activation_1022[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1026 (Conv2D)            (None, 35, 35, 96)   82944       activation_1025[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1027 (Conv2D)            (None, 35, 35, 64)   12288       average_pooling2d_6[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1021 (Batch (None, 35, 35, 96)   288         conv2d_1021[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1023 (Batch (None, 35, 35, 64)   192         conv2d_1023[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1026 (Batch (None, 35, 35, 96)   288         conv2d_1026[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1027 (Batch (None, 35, 35, 64)   192         conv2d_1027[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1021 (Activation)    (None, 35, 35, 96)   0           batch_normalization_1021[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1023 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1023[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1026 (Activation)    (None, 35, 35, 96)   0           batch_normalization_1026[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1027 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1027[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "mixed_5b (Concatenate)          (None, 35, 35, 320)  0           activation_1021[0][0]            \n",
      "                                                                 activation_1023[0][0]            \n",
      "                                                                 activation_1026[0][0]            \n",
      "                                                                 activation_1027[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1031 (Conv2D)            (None, 35, 35, 32)   10240       mixed_5b[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1031 (Batch (None, 35, 35, 32)   96          conv2d_1031[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1031 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1031[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1029 (Conv2D)            (None, 35, 35, 32)   10240       mixed_5b[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1032 (Conv2D)            (None, 35, 35, 48)   13824       activation_1031[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1029 (Batch (None, 35, 35, 32)   96          conv2d_1029[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1032 (Batch (None, 35, 35, 48)   144         conv2d_1032[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1029 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1029[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1032 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1032[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1028 (Conv2D)            (None, 35, 35, 32)   10240       mixed_5b[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1030 (Conv2D)            (None, 35, 35, 32)   9216        activation_1029[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1033 (Conv2D)            (None, 35, 35, 64)   27648       activation_1032[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1028 (Batch (None, 35, 35, 32)   96          conv2d_1028[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1030 (Batch (None, 35, 35, 32)   96          conv2d_1030[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1033 (Batch (None, 35, 35, 64)   192         conv2d_1033[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1028 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1028[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1030 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1030[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1033 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1033[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block35_1_mixed (Concatenate)   (None, 35, 35, 128)  0           activation_1028[0][0]            \n",
      "                                                                 activation_1030[0][0]            \n",
      "                                                                 activation_1033[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_1_conv (Conv2D)         (None, 35, 35, 320)  41280       block35_1_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_1 (Lambda)              (None, 35, 35, 320)  0           mixed_5b[0][0]                   \n",
      "                                                                 block35_1_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block35_1_ac (Activation)       (None, 35, 35, 320)  0           block35_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1037 (Conv2D)            (None, 35, 35, 32)   10240       block35_1_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1037 (Batch (None, 35, 35, 32)   96          conv2d_1037[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1037 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1037[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1035 (Conv2D)            (None, 35, 35, 32)   10240       block35_1_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1038 (Conv2D)            (None, 35, 35, 48)   13824       activation_1037[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1035 (Batch (None, 35, 35, 32)   96          conv2d_1035[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1038 (Batch (None, 35, 35, 48)   144         conv2d_1038[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1035 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1035[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1038 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1038[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1034 (Conv2D)            (None, 35, 35, 32)   10240       block35_1_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1036 (Conv2D)            (None, 35, 35, 32)   9216        activation_1035[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1039 (Conv2D)            (None, 35, 35, 64)   27648       activation_1038[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1034 (Batch (None, 35, 35, 32)   96          conv2d_1034[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1036 (Batch (None, 35, 35, 32)   96          conv2d_1036[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1039 (Batch (None, 35, 35, 64)   192         conv2d_1039[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1034 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1034[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1036 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1036[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1039 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1039[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block35_2_mixed (Concatenate)   (None, 35, 35, 128)  0           activation_1034[0][0]            \n",
      "                                                                 activation_1036[0][0]            \n",
      "                                                                 activation_1039[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_2_conv (Conv2D)         (None, 35, 35, 320)  41280       block35_2_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_2 (Lambda)              (None, 35, 35, 320)  0           block35_1_ac[0][0]               \n",
      "                                                                 block35_2_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block35_2_ac (Activation)       (None, 35, 35, 320)  0           block35_2[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1043 (Conv2D)            (None, 35, 35, 32)   10240       block35_2_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1043 (Batch (None, 35, 35, 32)   96          conv2d_1043[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1043 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1043[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1041 (Conv2D)            (None, 35, 35, 32)   10240       block35_2_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1044 (Conv2D)            (None, 35, 35, 48)   13824       activation_1043[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1041 (Batch (None, 35, 35, 32)   96          conv2d_1041[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1044 (Batch (None, 35, 35, 48)   144         conv2d_1044[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1041 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1041[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1044 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1044[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1040 (Conv2D)            (None, 35, 35, 32)   10240       block35_2_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1042 (Conv2D)            (None, 35, 35, 32)   9216        activation_1041[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1045 (Conv2D)            (None, 35, 35, 64)   27648       activation_1044[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1040 (Batch (None, 35, 35, 32)   96          conv2d_1040[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1042 (Batch (None, 35, 35, 32)   96          conv2d_1042[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1045 (Batch (None, 35, 35, 64)   192         conv2d_1045[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1040 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1040[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1042 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1042[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1045 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1045[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block35_3_mixed (Concatenate)   (None, 35, 35, 128)  0           activation_1040[0][0]            \n",
      "                                                                 activation_1042[0][0]            \n",
      "                                                                 activation_1045[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_3_conv (Conv2D)         (None, 35, 35, 320)  41280       block35_3_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_3 (Lambda)              (None, 35, 35, 320)  0           block35_2_ac[0][0]               \n",
      "                                                                 block35_3_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block35_3_ac (Activation)       (None, 35, 35, 320)  0           block35_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1049 (Conv2D)            (None, 35, 35, 32)   10240       block35_3_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1049 (Batch (None, 35, 35, 32)   96          conv2d_1049[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1049 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1049[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1047 (Conv2D)            (None, 35, 35, 32)   10240       block35_3_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1050 (Conv2D)            (None, 35, 35, 48)   13824       activation_1049[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1047 (Batch (None, 35, 35, 32)   96          conv2d_1047[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1050 (Batch (None, 35, 35, 48)   144         conv2d_1050[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1047 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1047[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1050 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1050[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1046 (Conv2D)            (None, 35, 35, 32)   10240       block35_3_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1048 (Conv2D)            (None, 35, 35, 32)   9216        activation_1047[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1051 (Conv2D)            (None, 35, 35, 64)   27648       activation_1050[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1046 (Batch (None, 35, 35, 32)   96          conv2d_1046[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1048 (Batch (None, 35, 35, 32)   96          conv2d_1048[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1051 (Batch (None, 35, 35, 64)   192         conv2d_1051[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1046 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1046[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1048 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1048[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1051 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1051[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block35_4_mixed (Concatenate)   (None, 35, 35, 128)  0           activation_1046[0][0]            \n",
      "                                                                 activation_1048[0][0]            \n",
      "                                                                 activation_1051[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_4_conv (Conv2D)         (None, 35, 35, 320)  41280       block35_4_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_4 (Lambda)              (None, 35, 35, 320)  0           block35_3_ac[0][0]               \n",
      "                                                                 block35_4_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block35_4_ac (Activation)       (None, 35, 35, 320)  0           block35_4[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1055 (Conv2D)            (None, 35, 35, 32)   10240       block35_4_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1055 (Batch (None, 35, 35, 32)   96          conv2d_1055[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1055 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1055[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1053 (Conv2D)            (None, 35, 35, 32)   10240       block35_4_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1056 (Conv2D)            (None, 35, 35, 48)   13824       activation_1055[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1053 (Batch (None, 35, 35, 32)   96          conv2d_1053[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1056 (Batch (None, 35, 35, 48)   144         conv2d_1056[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1053 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1053[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1056 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1056[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1052 (Conv2D)            (None, 35, 35, 32)   10240       block35_4_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1054 (Conv2D)            (None, 35, 35, 32)   9216        activation_1053[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1057 (Conv2D)            (None, 35, 35, 64)   27648       activation_1056[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1052 (Batch (None, 35, 35, 32)   96          conv2d_1052[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1054 (Batch (None, 35, 35, 32)   96          conv2d_1054[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1057 (Batch (None, 35, 35, 64)   192         conv2d_1057[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1052 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1052[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1054 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1054[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1057 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1057[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block35_5_mixed (Concatenate)   (None, 35, 35, 128)  0           activation_1052[0][0]            \n",
      "                                                                 activation_1054[0][0]            \n",
      "                                                                 activation_1057[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_5_conv (Conv2D)         (None, 35, 35, 320)  41280       block35_5_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_5 (Lambda)              (None, 35, 35, 320)  0           block35_4_ac[0][0]               \n",
      "                                                                 block35_5_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block35_5_ac (Activation)       (None, 35, 35, 320)  0           block35_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1061 (Conv2D)            (None, 35, 35, 32)   10240       block35_5_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1061 (Batch (None, 35, 35, 32)   96          conv2d_1061[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1061 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1061[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1059 (Conv2D)            (None, 35, 35, 32)   10240       block35_5_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1062 (Conv2D)            (None, 35, 35, 48)   13824       activation_1061[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1059 (Batch (None, 35, 35, 32)   96          conv2d_1059[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1062 (Batch (None, 35, 35, 48)   144         conv2d_1062[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1059 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1059[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1062 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1062[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1058 (Conv2D)            (None, 35, 35, 32)   10240       block35_5_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1060 (Conv2D)            (None, 35, 35, 32)   9216        activation_1059[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1063 (Conv2D)            (None, 35, 35, 64)   27648       activation_1062[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1058 (Batch (None, 35, 35, 32)   96          conv2d_1058[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1060 (Batch (None, 35, 35, 32)   96          conv2d_1060[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1063 (Batch (None, 35, 35, 64)   192         conv2d_1063[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1058 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1058[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1060 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1060[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1063 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1063[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block35_6_mixed (Concatenate)   (None, 35, 35, 128)  0           activation_1058[0][0]            \n",
      "                                                                 activation_1060[0][0]            \n",
      "                                                                 activation_1063[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_6_conv (Conv2D)         (None, 35, 35, 320)  41280       block35_6_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_6 (Lambda)              (None, 35, 35, 320)  0           block35_5_ac[0][0]               \n",
      "                                                                 block35_6_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block35_6_ac (Activation)       (None, 35, 35, 320)  0           block35_6[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1067 (Conv2D)            (None, 35, 35, 32)   10240       block35_6_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1067 (Batch (None, 35, 35, 32)   96          conv2d_1067[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1067 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1067[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1065 (Conv2D)            (None, 35, 35, 32)   10240       block35_6_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1068 (Conv2D)            (None, 35, 35, 48)   13824       activation_1067[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1065 (Batch (None, 35, 35, 32)   96          conv2d_1065[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1068 (Batch (None, 35, 35, 48)   144         conv2d_1068[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1065 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1065[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1068 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1068[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1064 (Conv2D)            (None, 35, 35, 32)   10240       block35_6_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1066 (Conv2D)            (None, 35, 35, 32)   9216        activation_1065[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1069 (Conv2D)            (None, 35, 35, 64)   27648       activation_1068[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1064 (Batch (None, 35, 35, 32)   96          conv2d_1064[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1066 (Batch (None, 35, 35, 32)   96          conv2d_1066[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1069 (Batch (None, 35, 35, 64)   192         conv2d_1069[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1064 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1064[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1066 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1066[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1069 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1069[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block35_7_mixed (Concatenate)   (None, 35, 35, 128)  0           activation_1064[0][0]            \n",
      "                                                                 activation_1066[0][0]            \n",
      "                                                                 activation_1069[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_7_conv (Conv2D)         (None, 35, 35, 320)  41280       block35_7_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_7 (Lambda)              (None, 35, 35, 320)  0           block35_6_ac[0][0]               \n",
      "                                                                 block35_7_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block35_7_ac (Activation)       (None, 35, 35, 320)  0           block35_7[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1073 (Conv2D)            (None, 35, 35, 32)   10240       block35_7_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1073 (Batch (None, 35, 35, 32)   96          conv2d_1073[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1073 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1073[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1071 (Conv2D)            (None, 35, 35, 32)   10240       block35_7_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1074 (Conv2D)            (None, 35, 35, 48)   13824       activation_1073[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1071 (Batch (None, 35, 35, 32)   96          conv2d_1071[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1074 (Batch (None, 35, 35, 48)   144         conv2d_1074[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1071 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1071[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1074 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1074[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1070 (Conv2D)            (None, 35, 35, 32)   10240       block35_7_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1072 (Conv2D)            (None, 35, 35, 32)   9216        activation_1071[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1075 (Conv2D)            (None, 35, 35, 64)   27648       activation_1074[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1070 (Batch (None, 35, 35, 32)   96          conv2d_1070[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1072 (Batch (None, 35, 35, 32)   96          conv2d_1072[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1075 (Batch (None, 35, 35, 64)   192         conv2d_1075[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1070 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1070[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1072 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1072[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1075 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1075[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block35_8_mixed (Concatenate)   (None, 35, 35, 128)  0           activation_1070[0][0]            \n",
      "                                                                 activation_1072[0][0]            \n",
      "                                                                 activation_1075[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_8_conv (Conv2D)         (None, 35, 35, 320)  41280       block35_8_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_8 (Lambda)              (None, 35, 35, 320)  0           block35_7_ac[0][0]               \n",
      "                                                                 block35_8_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block35_8_ac (Activation)       (None, 35, 35, 320)  0           block35_8[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1079 (Conv2D)            (None, 35, 35, 32)   10240       block35_8_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1079 (Batch (None, 35, 35, 32)   96          conv2d_1079[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1079 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1079[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1077 (Conv2D)            (None, 35, 35, 32)   10240       block35_8_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1080 (Conv2D)            (None, 35, 35, 48)   13824       activation_1079[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1077 (Batch (None, 35, 35, 32)   96          conv2d_1077[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1080 (Batch (None, 35, 35, 48)   144         conv2d_1080[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1077 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1077[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1080 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1080[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1076 (Conv2D)            (None, 35, 35, 32)   10240       block35_8_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1078 (Conv2D)            (None, 35, 35, 32)   9216        activation_1077[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1081 (Conv2D)            (None, 35, 35, 64)   27648       activation_1080[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1076 (Batch (None, 35, 35, 32)   96          conv2d_1076[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1078 (Batch (None, 35, 35, 32)   96          conv2d_1078[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1081 (Batch (None, 35, 35, 64)   192         conv2d_1081[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1076 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1076[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1078 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1078[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1081 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1081[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block35_9_mixed (Concatenate)   (None, 35, 35, 128)  0           activation_1076[0][0]            \n",
      "                                                                 activation_1078[0][0]            \n",
      "                                                                 activation_1081[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_9_conv (Conv2D)         (None, 35, 35, 320)  41280       block35_9_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_9 (Lambda)              (None, 35, 35, 320)  0           block35_8_ac[0][0]               \n",
      "                                                                 block35_9_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block35_9_ac (Activation)       (None, 35, 35, 320)  0           block35_9[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1085 (Conv2D)            (None, 35, 35, 32)   10240       block35_9_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1085 (Batch (None, 35, 35, 32)   96          conv2d_1085[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1085 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1085[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1083 (Conv2D)            (None, 35, 35, 32)   10240       block35_9_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1086 (Conv2D)            (None, 35, 35, 48)   13824       activation_1085[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1083 (Batch (None, 35, 35, 32)   96          conv2d_1083[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1086 (Batch (None, 35, 35, 48)   144         conv2d_1086[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1083 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1083[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1086 (Activation)    (None, 35, 35, 48)   0           batch_normalization_1086[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1082 (Conv2D)            (None, 35, 35, 32)   10240       block35_9_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1084 (Conv2D)            (None, 35, 35, 32)   9216        activation_1083[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1087 (Conv2D)            (None, 35, 35, 64)   27648       activation_1086[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1082 (Batch (None, 35, 35, 32)   96          conv2d_1082[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1084 (Batch (None, 35, 35, 32)   96          conv2d_1084[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1087 (Batch (None, 35, 35, 64)   192         conv2d_1087[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1082 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1082[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1084 (Activation)    (None, 35, 35, 32)   0           batch_normalization_1084[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1087 (Activation)    (None, 35, 35, 64)   0           batch_normalization_1087[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block35_10_mixed (Concatenate)  (None, 35, 35, 128)  0           activation_1082[0][0]            \n",
      "                                                                 activation_1084[0][0]            \n",
      "                                                                 activation_1087[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_10_conv (Conv2D)        (None, 35, 35, 320)  41280       block35_10_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block35_10 (Lambda)             (None, 35, 35, 320)  0           block35_9_ac[0][0]               \n",
      "                                                                 block35_10_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block35_10_ac (Activation)      (None, 35, 35, 320)  0           block35_10[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1089 (Conv2D)            (None, 35, 35, 256)  81920       block35_10_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1089 (Batch (None, 35, 35, 256)  768         conv2d_1089[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1089 (Activation)    (None, 35, 35, 256)  0           batch_normalization_1089[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1090 (Conv2D)            (None, 35, 35, 256)  589824      activation_1089[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1090 (Batch (None, 35, 35, 256)  768         conv2d_1090[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1090 (Activation)    (None, 35, 35, 256)  0           batch_normalization_1090[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1088 (Conv2D)            (None, 17, 17, 384)  1105920     block35_10_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1091 (Conv2D)            (None, 17, 17, 384)  884736      activation_1090[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1088 (Batch (None, 17, 17, 384)  1152        conv2d_1088[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1091 (Batch (None, 17, 17, 384)  1152        conv2d_1091[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1088 (Activation)    (None, 17, 17, 384)  0           batch_normalization_1088[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1091 (Activation)    (None, 17, 17, 384)  0           batch_normalization_1091[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_23 (MaxPooling2D) (None, 17, 17, 320)  0           block35_10_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "mixed_6a (Concatenate)          (None, 17, 17, 1088) 0           activation_1088[0][0]            \n",
      "                                                                 activation_1091[0][0]            \n",
      "                                                                 max_pooling2d_23[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1093 (Conv2D)            (None, 17, 17, 128)  139264      mixed_6a[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1093 (Batch (None, 17, 17, 128)  384         conv2d_1093[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1093 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1093[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1094 (Conv2D)            (None, 17, 17, 160)  143360      activation_1093[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1094 (Batch (None, 17, 17, 160)  480         conv2d_1094[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1094 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1094[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1092 (Conv2D)            (None, 17, 17, 192)  208896      mixed_6a[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1095 (Conv2D)            (None, 17, 17, 192)  215040      activation_1094[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1092 (Batch (None, 17, 17, 192)  576         conv2d_1092[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1095 (Batch (None, 17, 17, 192)  576         conv2d_1095[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1092 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1092[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1095 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1095[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_1_mixed (Concatenate)   (None, 17, 17, 384)  0           activation_1092[0][0]            \n",
      "                                                                 activation_1095[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_1_conv (Conv2D)         (None, 17, 17, 1088) 418880      block17_1_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_1 (Lambda)              (None, 17, 17, 1088) 0           mixed_6a[0][0]                   \n",
      "                                                                 block17_1_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block17_1_ac (Activation)       (None, 17, 17, 1088) 0           block17_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1097 (Conv2D)            (None, 17, 17, 128)  139264      block17_1_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1097 (Batch (None, 17, 17, 128)  384         conv2d_1097[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1097 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1097[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1098 (Conv2D)            (None, 17, 17, 160)  143360      activation_1097[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1098 (Batch (None, 17, 17, 160)  480         conv2d_1098[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1098 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1098[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1096 (Conv2D)            (None, 17, 17, 192)  208896      block17_1_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1099 (Conv2D)            (None, 17, 17, 192)  215040      activation_1098[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1096 (Batch (None, 17, 17, 192)  576         conv2d_1096[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1099 (Batch (None, 17, 17, 192)  576         conv2d_1099[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1096 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1096[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1099 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1099[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_2_mixed (Concatenate)   (None, 17, 17, 384)  0           activation_1096[0][0]            \n",
      "                                                                 activation_1099[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_2_conv (Conv2D)         (None, 17, 17, 1088) 418880      block17_2_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_2 (Lambda)              (None, 17, 17, 1088) 0           block17_1_ac[0][0]               \n",
      "                                                                 block17_2_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block17_2_ac (Activation)       (None, 17, 17, 1088) 0           block17_2[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1101 (Conv2D)            (None, 17, 17, 128)  139264      block17_2_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1101 (Batch (None, 17, 17, 128)  384         conv2d_1101[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1101 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1101[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1102 (Conv2D)            (None, 17, 17, 160)  143360      activation_1101[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1102 (Batch (None, 17, 17, 160)  480         conv2d_1102[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1102 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1102[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1100 (Conv2D)            (None, 17, 17, 192)  208896      block17_2_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1103 (Conv2D)            (None, 17, 17, 192)  215040      activation_1102[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1100 (Batch (None, 17, 17, 192)  576         conv2d_1100[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1103 (Batch (None, 17, 17, 192)  576         conv2d_1103[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1100 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1100[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1103 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1103[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_3_mixed (Concatenate)   (None, 17, 17, 384)  0           activation_1100[0][0]            \n",
      "                                                                 activation_1103[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_3_conv (Conv2D)         (None, 17, 17, 1088) 418880      block17_3_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_3 (Lambda)              (None, 17, 17, 1088) 0           block17_2_ac[0][0]               \n",
      "                                                                 block17_3_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block17_3_ac (Activation)       (None, 17, 17, 1088) 0           block17_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1105 (Conv2D)            (None, 17, 17, 128)  139264      block17_3_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1105 (Batch (None, 17, 17, 128)  384         conv2d_1105[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1105 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1105[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1106 (Conv2D)            (None, 17, 17, 160)  143360      activation_1105[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1106 (Batch (None, 17, 17, 160)  480         conv2d_1106[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1106 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1106[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1104 (Conv2D)            (None, 17, 17, 192)  208896      block17_3_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1107 (Conv2D)            (None, 17, 17, 192)  215040      activation_1106[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1104 (Batch (None, 17, 17, 192)  576         conv2d_1104[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1107 (Batch (None, 17, 17, 192)  576         conv2d_1107[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1104 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1104[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1107 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1107[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_4_mixed (Concatenate)   (None, 17, 17, 384)  0           activation_1104[0][0]            \n",
      "                                                                 activation_1107[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_4_conv (Conv2D)         (None, 17, 17, 1088) 418880      block17_4_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_4 (Lambda)              (None, 17, 17, 1088) 0           block17_3_ac[0][0]               \n",
      "                                                                 block17_4_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block17_4_ac (Activation)       (None, 17, 17, 1088) 0           block17_4[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1109 (Conv2D)            (None, 17, 17, 128)  139264      block17_4_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1109 (Batch (None, 17, 17, 128)  384         conv2d_1109[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1109 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1109[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1110 (Conv2D)            (None, 17, 17, 160)  143360      activation_1109[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1110 (Batch (None, 17, 17, 160)  480         conv2d_1110[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1110 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1110[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1108 (Conv2D)            (None, 17, 17, 192)  208896      block17_4_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1111 (Conv2D)            (None, 17, 17, 192)  215040      activation_1110[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1108 (Batch (None, 17, 17, 192)  576         conv2d_1108[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1111 (Batch (None, 17, 17, 192)  576         conv2d_1111[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1108 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1108[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1111 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1111[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_5_mixed (Concatenate)   (None, 17, 17, 384)  0           activation_1108[0][0]            \n",
      "                                                                 activation_1111[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_5_conv (Conv2D)         (None, 17, 17, 1088) 418880      block17_5_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_5 (Lambda)              (None, 17, 17, 1088) 0           block17_4_ac[0][0]               \n",
      "                                                                 block17_5_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block17_5_ac (Activation)       (None, 17, 17, 1088) 0           block17_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1113 (Conv2D)            (None, 17, 17, 128)  139264      block17_5_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1113 (Batch (None, 17, 17, 128)  384         conv2d_1113[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1113 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1113[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1114 (Conv2D)            (None, 17, 17, 160)  143360      activation_1113[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1114 (Batch (None, 17, 17, 160)  480         conv2d_1114[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1114 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1114[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1112 (Conv2D)            (None, 17, 17, 192)  208896      block17_5_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1115 (Conv2D)            (None, 17, 17, 192)  215040      activation_1114[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1112 (Batch (None, 17, 17, 192)  576         conv2d_1112[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1115 (Batch (None, 17, 17, 192)  576         conv2d_1115[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1112 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1112[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1115 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1115[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_6_mixed (Concatenate)   (None, 17, 17, 384)  0           activation_1112[0][0]            \n",
      "                                                                 activation_1115[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_6_conv (Conv2D)         (None, 17, 17, 1088) 418880      block17_6_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_6 (Lambda)              (None, 17, 17, 1088) 0           block17_5_ac[0][0]               \n",
      "                                                                 block17_6_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block17_6_ac (Activation)       (None, 17, 17, 1088) 0           block17_6[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1117 (Conv2D)            (None, 17, 17, 128)  139264      block17_6_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1117 (Batch (None, 17, 17, 128)  384         conv2d_1117[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1117 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1117[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1118 (Conv2D)            (None, 17, 17, 160)  143360      activation_1117[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1118 (Batch (None, 17, 17, 160)  480         conv2d_1118[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1118 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1118[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1116 (Conv2D)            (None, 17, 17, 192)  208896      block17_6_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1119 (Conv2D)            (None, 17, 17, 192)  215040      activation_1118[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1116 (Batch (None, 17, 17, 192)  576         conv2d_1116[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1119 (Batch (None, 17, 17, 192)  576         conv2d_1119[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1116 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1116[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1119 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1119[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_7_mixed (Concatenate)   (None, 17, 17, 384)  0           activation_1116[0][0]            \n",
      "                                                                 activation_1119[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_7_conv (Conv2D)         (None, 17, 17, 1088) 418880      block17_7_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_7 (Lambda)              (None, 17, 17, 1088) 0           block17_6_ac[0][0]               \n",
      "                                                                 block17_7_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block17_7_ac (Activation)       (None, 17, 17, 1088) 0           block17_7[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1121 (Conv2D)            (None, 17, 17, 128)  139264      block17_7_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1121 (Batch (None, 17, 17, 128)  384         conv2d_1121[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1121 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1121[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1122 (Conv2D)            (None, 17, 17, 160)  143360      activation_1121[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1122 (Batch (None, 17, 17, 160)  480         conv2d_1122[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1122 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1122[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1120 (Conv2D)            (None, 17, 17, 192)  208896      block17_7_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1123 (Conv2D)            (None, 17, 17, 192)  215040      activation_1122[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1120 (Batch (None, 17, 17, 192)  576         conv2d_1120[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1123 (Batch (None, 17, 17, 192)  576         conv2d_1123[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1120 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1120[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1123 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1123[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_8_mixed (Concatenate)   (None, 17, 17, 384)  0           activation_1120[0][0]            \n",
      "                                                                 activation_1123[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_8_conv (Conv2D)         (None, 17, 17, 1088) 418880      block17_8_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_8 (Lambda)              (None, 17, 17, 1088) 0           block17_7_ac[0][0]               \n",
      "                                                                 block17_8_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block17_8_ac (Activation)       (None, 17, 17, 1088) 0           block17_8[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1125 (Conv2D)            (None, 17, 17, 128)  139264      block17_8_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1125 (Batch (None, 17, 17, 128)  384         conv2d_1125[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1125 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1125[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1126 (Conv2D)            (None, 17, 17, 160)  143360      activation_1125[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1126 (Batch (None, 17, 17, 160)  480         conv2d_1126[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1126 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1126[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1124 (Conv2D)            (None, 17, 17, 192)  208896      block17_8_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1127 (Conv2D)            (None, 17, 17, 192)  215040      activation_1126[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1124 (Batch (None, 17, 17, 192)  576         conv2d_1124[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1127 (Batch (None, 17, 17, 192)  576         conv2d_1127[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1124 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1124[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1127 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1127[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_9_mixed (Concatenate)   (None, 17, 17, 384)  0           activation_1124[0][0]            \n",
      "                                                                 activation_1127[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_9_conv (Conv2D)         (None, 17, 17, 1088) 418880      block17_9_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_9 (Lambda)              (None, 17, 17, 1088) 0           block17_8_ac[0][0]               \n",
      "                                                                 block17_9_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block17_9_ac (Activation)       (None, 17, 17, 1088) 0           block17_9[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1129 (Conv2D)            (None, 17, 17, 128)  139264      block17_9_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1129 (Batch (None, 17, 17, 128)  384         conv2d_1129[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1129 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1129[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1130 (Conv2D)            (None, 17, 17, 160)  143360      activation_1129[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1130 (Batch (None, 17, 17, 160)  480         conv2d_1130[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1130 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1130[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1128 (Conv2D)            (None, 17, 17, 192)  208896      block17_9_ac[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1131 (Conv2D)            (None, 17, 17, 192)  215040      activation_1130[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1128 (Batch (None, 17, 17, 192)  576         conv2d_1128[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1131 (Batch (None, 17, 17, 192)  576         conv2d_1131[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1128 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1128[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1131 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1131[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_10_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1128[0][0]            \n",
      "                                                                 activation_1131[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_10_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_10_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_10 (Lambda)             (None, 17, 17, 1088) 0           block17_9_ac[0][0]               \n",
      "                                                                 block17_10_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_10_ac (Activation)      (None, 17, 17, 1088) 0           block17_10[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1133 (Conv2D)            (None, 17, 17, 128)  139264      block17_10_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1133 (Batch (None, 17, 17, 128)  384         conv2d_1133[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1133 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1133[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1134 (Conv2D)            (None, 17, 17, 160)  143360      activation_1133[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1134 (Batch (None, 17, 17, 160)  480         conv2d_1134[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1134 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1134[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1132 (Conv2D)            (None, 17, 17, 192)  208896      block17_10_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1135 (Conv2D)            (None, 17, 17, 192)  215040      activation_1134[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1132 (Batch (None, 17, 17, 192)  576         conv2d_1132[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1135 (Batch (None, 17, 17, 192)  576         conv2d_1135[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1132 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1132[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1135 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1135[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_11_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1132[0][0]            \n",
      "                                                                 activation_1135[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_11_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_11_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_11 (Lambda)             (None, 17, 17, 1088) 0           block17_10_ac[0][0]              \n",
      "                                                                 block17_11_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_11_ac (Activation)      (None, 17, 17, 1088) 0           block17_11[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1137 (Conv2D)            (None, 17, 17, 128)  139264      block17_11_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1137 (Batch (None, 17, 17, 128)  384         conv2d_1137[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1137 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1137[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1138 (Conv2D)            (None, 17, 17, 160)  143360      activation_1137[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1138 (Batch (None, 17, 17, 160)  480         conv2d_1138[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1138 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1138[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1136 (Conv2D)            (None, 17, 17, 192)  208896      block17_11_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1139 (Conv2D)            (None, 17, 17, 192)  215040      activation_1138[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1136 (Batch (None, 17, 17, 192)  576         conv2d_1136[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1139 (Batch (None, 17, 17, 192)  576         conv2d_1139[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1136 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1136[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1139 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1139[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_12_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1136[0][0]            \n",
      "                                                                 activation_1139[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_12_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_12_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_12 (Lambda)             (None, 17, 17, 1088) 0           block17_11_ac[0][0]              \n",
      "                                                                 block17_12_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_12_ac (Activation)      (None, 17, 17, 1088) 0           block17_12[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1141 (Conv2D)            (None, 17, 17, 128)  139264      block17_12_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1141 (Batch (None, 17, 17, 128)  384         conv2d_1141[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1141 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1141[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1142 (Conv2D)            (None, 17, 17, 160)  143360      activation_1141[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1142 (Batch (None, 17, 17, 160)  480         conv2d_1142[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1142 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1142[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1140 (Conv2D)            (None, 17, 17, 192)  208896      block17_12_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1143 (Conv2D)            (None, 17, 17, 192)  215040      activation_1142[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1140 (Batch (None, 17, 17, 192)  576         conv2d_1140[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1143 (Batch (None, 17, 17, 192)  576         conv2d_1143[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1140 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1140[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1143 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1143[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_13_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1140[0][0]            \n",
      "                                                                 activation_1143[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_13_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_13_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_13 (Lambda)             (None, 17, 17, 1088) 0           block17_12_ac[0][0]              \n",
      "                                                                 block17_13_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_13_ac (Activation)      (None, 17, 17, 1088) 0           block17_13[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1145 (Conv2D)            (None, 17, 17, 128)  139264      block17_13_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1145 (Batch (None, 17, 17, 128)  384         conv2d_1145[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1145 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1145[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1146 (Conv2D)            (None, 17, 17, 160)  143360      activation_1145[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1146 (Batch (None, 17, 17, 160)  480         conv2d_1146[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1146 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1146[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1144 (Conv2D)            (None, 17, 17, 192)  208896      block17_13_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1147 (Conv2D)            (None, 17, 17, 192)  215040      activation_1146[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1144 (Batch (None, 17, 17, 192)  576         conv2d_1144[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1147 (Batch (None, 17, 17, 192)  576         conv2d_1147[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1144 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1144[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1147 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1147[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_14_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1144[0][0]            \n",
      "                                                                 activation_1147[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_14_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_14_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_14 (Lambda)             (None, 17, 17, 1088) 0           block17_13_ac[0][0]              \n",
      "                                                                 block17_14_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_14_ac (Activation)      (None, 17, 17, 1088) 0           block17_14[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1149 (Conv2D)            (None, 17, 17, 128)  139264      block17_14_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1149 (Batch (None, 17, 17, 128)  384         conv2d_1149[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1149 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1149[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1150 (Conv2D)            (None, 17, 17, 160)  143360      activation_1149[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1150 (Batch (None, 17, 17, 160)  480         conv2d_1150[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1150 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1150[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1148 (Conv2D)            (None, 17, 17, 192)  208896      block17_14_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1151 (Conv2D)            (None, 17, 17, 192)  215040      activation_1150[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1148 (Batch (None, 17, 17, 192)  576         conv2d_1148[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1151 (Batch (None, 17, 17, 192)  576         conv2d_1151[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1148 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1148[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1151 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1151[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_15_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1148[0][0]            \n",
      "                                                                 activation_1151[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_15_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_15_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_15 (Lambda)             (None, 17, 17, 1088) 0           block17_14_ac[0][0]              \n",
      "                                                                 block17_15_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_15_ac (Activation)      (None, 17, 17, 1088) 0           block17_15[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1153 (Conv2D)            (None, 17, 17, 128)  139264      block17_15_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1153 (Batch (None, 17, 17, 128)  384         conv2d_1153[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1153 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1153[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1154 (Conv2D)            (None, 17, 17, 160)  143360      activation_1153[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1154 (Batch (None, 17, 17, 160)  480         conv2d_1154[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1154 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1154[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1152 (Conv2D)            (None, 17, 17, 192)  208896      block17_15_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1155 (Conv2D)            (None, 17, 17, 192)  215040      activation_1154[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1152 (Batch (None, 17, 17, 192)  576         conv2d_1152[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1155 (Batch (None, 17, 17, 192)  576         conv2d_1155[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1152 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1152[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1155 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1155[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_16_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1152[0][0]            \n",
      "                                                                 activation_1155[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_16_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_16_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_16 (Lambda)             (None, 17, 17, 1088) 0           block17_15_ac[0][0]              \n",
      "                                                                 block17_16_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_16_ac (Activation)      (None, 17, 17, 1088) 0           block17_16[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1157 (Conv2D)            (None, 17, 17, 128)  139264      block17_16_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1157 (Batch (None, 17, 17, 128)  384         conv2d_1157[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1157 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1157[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1158 (Conv2D)            (None, 17, 17, 160)  143360      activation_1157[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1158 (Batch (None, 17, 17, 160)  480         conv2d_1158[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1158 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1158[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1156 (Conv2D)            (None, 17, 17, 192)  208896      block17_16_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1159 (Conv2D)            (None, 17, 17, 192)  215040      activation_1158[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1156 (Batch (None, 17, 17, 192)  576         conv2d_1156[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1159 (Batch (None, 17, 17, 192)  576         conv2d_1159[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1156 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1156[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1159 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1159[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_17_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1156[0][0]            \n",
      "                                                                 activation_1159[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_17_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_17_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_17 (Lambda)             (None, 17, 17, 1088) 0           block17_16_ac[0][0]              \n",
      "                                                                 block17_17_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_17_ac (Activation)      (None, 17, 17, 1088) 0           block17_17[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1161 (Conv2D)            (None, 17, 17, 128)  139264      block17_17_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1161 (Batch (None, 17, 17, 128)  384         conv2d_1161[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1161 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1161[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1162 (Conv2D)            (None, 17, 17, 160)  143360      activation_1161[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1162 (Batch (None, 17, 17, 160)  480         conv2d_1162[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1162 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1162[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1160 (Conv2D)            (None, 17, 17, 192)  208896      block17_17_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1163 (Conv2D)            (None, 17, 17, 192)  215040      activation_1162[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1160 (Batch (None, 17, 17, 192)  576         conv2d_1160[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1163 (Batch (None, 17, 17, 192)  576         conv2d_1163[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1160 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1160[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1163 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1163[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_18_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1160[0][0]            \n",
      "                                                                 activation_1163[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_18_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_18_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_18 (Lambda)             (None, 17, 17, 1088) 0           block17_17_ac[0][0]              \n",
      "                                                                 block17_18_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_18_ac (Activation)      (None, 17, 17, 1088) 0           block17_18[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1165 (Conv2D)            (None, 17, 17, 128)  139264      block17_18_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1165 (Batch (None, 17, 17, 128)  384         conv2d_1165[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1165 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1165[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1166 (Conv2D)            (None, 17, 17, 160)  143360      activation_1165[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1166 (Batch (None, 17, 17, 160)  480         conv2d_1166[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1166 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1166[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1164 (Conv2D)            (None, 17, 17, 192)  208896      block17_18_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1167 (Conv2D)            (None, 17, 17, 192)  215040      activation_1166[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1164 (Batch (None, 17, 17, 192)  576         conv2d_1164[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1167 (Batch (None, 17, 17, 192)  576         conv2d_1167[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1164 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1164[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1167 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1167[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_19_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1164[0][0]            \n",
      "                                                                 activation_1167[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_19_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_19_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_19 (Lambda)             (None, 17, 17, 1088) 0           block17_18_ac[0][0]              \n",
      "                                                                 block17_19_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_19_ac (Activation)      (None, 17, 17, 1088) 0           block17_19[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1169 (Conv2D)            (None, 17, 17, 128)  139264      block17_19_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1169 (Batch (None, 17, 17, 128)  384         conv2d_1169[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1169 (Activation)    (None, 17, 17, 128)  0           batch_normalization_1169[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1170 (Conv2D)            (None, 17, 17, 160)  143360      activation_1169[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1170 (Batch (None, 17, 17, 160)  480         conv2d_1170[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1170 (Activation)    (None, 17, 17, 160)  0           batch_normalization_1170[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1168 (Conv2D)            (None, 17, 17, 192)  208896      block17_19_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1171 (Conv2D)            (None, 17, 17, 192)  215040      activation_1170[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1168 (Batch (None, 17, 17, 192)  576         conv2d_1168[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1171 (Batch (None, 17, 17, 192)  576         conv2d_1171[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1168 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1168[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1171 (Activation)    (None, 17, 17, 192)  0           batch_normalization_1171[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block17_20_mixed (Concatenate)  (None, 17, 17, 384)  0           activation_1168[0][0]            \n",
      "                                                                 activation_1171[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_20_conv (Conv2D)        (None, 17, 17, 1088) 418880      block17_20_mixed[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block17_20 (Lambda)             (None, 17, 17, 1088) 0           block17_19_ac[0][0]              \n",
      "                                                                 block17_20_conv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block17_20_ac (Activation)      (None, 17, 17, 1088) 0           block17_20[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1176 (Conv2D)            (None, 17, 17, 256)  278528      block17_20_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1176 (Batch (None, 17, 17, 256)  768         conv2d_1176[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1176 (Activation)    (None, 17, 17, 256)  0           batch_normalization_1176[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1172 (Conv2D)            (None, 17, 17, 256)  278528      block17_20_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1174 (Conv2D)            (None, 17, 17, 256)  278528      block17_20_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1177 (Conv2D)            (None, 17, 17, 288)  663552      activation_1176[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1172 (Batch (None, 17, 17, 256)  768         conv2d_1172[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1174 (Batch (None, 17, 17, 256)  768         conv2d_1174[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1177 (Batch (None, 17, 17, 288)  864         conv2d_1177[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1172 (Activation)    (None, 17, 17, 256)  0           batch_normalization_1172[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1174 (Activation)    (None, 17, 17, 256)  0           batch_normalization_1174[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1177 (Activation)    (None, 17, 17, 288)  0           batch_normalization_1177[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1173 (Conv2D)            (None, 8, 8, 384)    884736      activation_1172[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1175 (Conv2D)            (None, 8, 8, 288)    663552      activation_1174[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1178 (Conv2D)            (None, 8, 8, 320)    829440      activation_1177[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1173 (Batch (None, 8, 8, 384)    1152        conv2d_1173[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1175 (Batch (None, 8, 8, 288)    864         conv2d_1175[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1178 (Batch (None, 8, 8, 320)    960         conv2d_1178[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1173 (Activation)    (None, 8, 8, 384)    0           batch_normalization_1173[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1175 (Activation)    (None, 8, 8, 288)    0           batch_normalization_1175[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1178 (Activation)    (None, 8, 8, 320)    0           batch_normalization_1178[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_24 (MaxPooling2D) (None, 8, 8, 1088)   0           block17_20_ac[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "mixed_7a (Concatenate)          (None, 8, 8, 2080)   0           activation_1173[0][0]            \n",
      "                                                                 activation_1175[0][0]            \n",
      "                                                                 activation_1178[0][0]            \n",
      "                                                                 max_pooling2d_24[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1180 (Conv2D)            (None, 8, 8, 192)    399360      mixed_7a[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1180 (Batch (None, 8, 8, 192)    576         conv2d_1180[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1180 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1180[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1181 (Conv2D)            (None, 8, 8, 224)    129024      activation_1180[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1181 (Batch (None, 8, 8, 224)    672         conv2d_1181[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1181 (Activation)    (None, 8, 8, 224)    0           batch_normalization_1181[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1179 (Conv2D)            (None, 8, 8, 192)    399360      mixed_7a[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1182 (Conv2D)            (None, 8, 8, 256)    172032      activation_1181[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1179 (Batch (None, 8, 8, 192)    576         conv2d_1179[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1182 (Batch (None, 8, 8, 256)    768         conv2d_1182[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1179 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1179[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1182 (Activation)    (None, 8, 8, 256)    0           batch_normalization_1182[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block8_1_mixed (Concatenate)    (None, 8, 8, 448)    0           activation_1179[0][0]            \n",
      "                                                                 activation_1182[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_1_conv (Conv2D)          (None, 8, 8, 2080)   933920      block8_1_mixed[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block8_1 (Lambda)               (None, 8, 8, 2080)   0           mixed_7a[0][0]                   \n",
      "                                                                 block8_1_conv[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "block8_1_ac (Activation)        (None, 8, 8, 2080)   0           block8_1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1184 (Conv2D)            (None, 8, 8, 192)    399360      block8_1_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1184 (Batch (None, 8, 8, 192)    576         conv2d_1184[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1184 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1184[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1185 (Conv2D)            (None, 8, 8, 224)    129024      activation_1184[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1185 (Batch (None, 8, 8, 224)    672         conv2d_1185[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1185 (Activation)    (None, 8, 8, 224)    0           batch_normalization_1185[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1183 (Conv2D)            (None, 8, 8, 192)    399360      block8_1_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1186 (Conv2D)            (None, 8, 8, 256)    172032      activation_1185[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1183 (Batch (None, 8, 8, 192)    576         conv2d_1183[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1186 (Batch (None, 8, 8, 256)    768         conv2d_1186[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1183 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1183[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1186 (Activation)    (None, 8, 8, 256)    0           batch_normalization_1186[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block8_2_mixed (Concatenate)    (None, 8, 8, 448)    0           activation_1183[0][0]            \n",
      "                                                                 activation_1186[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_2_conv (Conv2D)          (None, 8, 8, 2080)   933920      block8_2_mixed[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block8_2 (Lambda)               (None, 8, 8, 2080)   0           block8_1_ac[0][0]                \n",
      "                                                                 block8_2_conv[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "block8_2_ac (Activation)        (None, 8, 8, 2080)   0           block8_2[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1188 (Conv2D)            (None, 8, 8, 192)    399360      block8_2_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1188 (Batch (None, 8, 8, 192)    576         conv2d_1188[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1188 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1188[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1189 (Conv2D)            (None, 8, 8, 224)    129024      activation_1188[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1189 (Batch (None, 8, 8, 224)    672         conv2d_1189[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1189 (Activation)    (None, 8, 8, 224)    0           batch_normalization_1189[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1187 (Conv2D)            (None, 8, 8, 192)    399360      block8_2_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1190 (Conv2D)            (None, 8, 8, 256)    172032      activation_1189[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1187 (Batch (None, 8, 8, 192)    576         conv2d_1187[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1190 (Batch (None, 8, 8, 256)    768         conv2d_1190[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1187 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1187[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1190 (Activation)    (None, 8, 8, 256)    0           batch_normalization_1190[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block8_3_mixed (Concatenate)    (None, 8, 8, 448)    0           activation_1187[0][0]            \n",
      "                                                                 activation_1190[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_3_conv (Conv2D)          (None, 8, 8, 2080)   933920      block8_3_mixed[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block8_3 (Lambda)               (None, 8, 8, 2080)   0           block8_2_ac[0][0]                \n",
      "                                                                 block8_3_conv[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "block8_3_ac (Activation)        (None, 8, 8, 2080)   0           block8_3[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1192 (Conv2D)            (None, 8, 8, 192)    399360      block8_3_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1192 (Batch (None, 8, 8, 192)    576         conv2d_1192[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1192 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1192[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1193 (Conv2D)            (None, 8, 8, 224)    129024      activation_1192[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1193 (Batch (None, 8, 8, 224)    672         conv2d_1193[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1193 (Activation)    (None, 8, 8, 224)    0           batch_normalization_1193[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1191 (Conv2D)            (None, 8, 8, 192)    399360      block8_3_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1194 (Conv2D)            (None, 8, 8, 256)    172032      activation_1193[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1191 (Batch (None, 8, 8, 192)    576         conv2d_1191[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1194 (Batch (None, 8, 8, 256)    768         conv2d_1194[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1191 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1191[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1194 (Activation)    (None, 8, 8, 256)    0           batch_normalization_1194[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block8_4_mixed (Concatenate)    (None, 8, 8, 448)    0           activation_1191[0][0]            \n",
      "                                                                 activation_1194[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_4_conv (Conv2D)          (None, 8, 8, 2080)   933920      block8_4_mixed[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block8_4 (Lambda)               (None, 8, 8, 2080)   0           block8_3_ac[0][0]                \n",
      "                                                                 block8_4_conv[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "block8_4_ac (Activation)        (None, 8, 8, 2080)   0           block8_4[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1196 (Conv2D)            (None, 8, 8, 192)    399360      block8_4_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1196 (Batch (None, 8, 8, 192)    576         conv2d_1196[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1196 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1196[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1197 (Conv2D)            (None, 8, 8, 224)    129024      activation_1196[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1197 (Batch (None, 8, 8, 224)    672         conv2d_1197[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1197 (Activation)    (None, 8, 8, 224)    0           batch_normalization_1197[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1195 (Conv2D)            (None, 8, 8, 192)    399360      block8_4_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1198 (Conv2D)            (None, 8, 8, 256)    172032      activation_1197[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1195 (Batch (None, 8, 8, 192)    576         conv2d_1195[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1198 (Batch (None, 8, 8, 256)    768         conv2d_1198[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1195 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1195[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1198 (Activation)    (None, 8, 8, 256)    0           batch_normalization_1198[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block8_5_mixed (Concatenate)    (None, 8, 8, 448)    0           activation_1195[0][0]            \n",
      "                                                                 activation_1198[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_5_conv (Conv2D)          (None, 8, 8, 2080)   933920      block8_5_mixed[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block8_5 (Lambda)               (None, 8, 8, 2080)   0           block8_4_ac[0][0]                \n",
      "                                                                 block8_5_conv[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "block8_5_ac (Activation)        (None, 8, 8, 2080)   0           block8_5[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1200 (Conv2D)            (None, 8, 8, 192)    399360      block8_5_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1200 (Batch (None, 8, 8, 192)    576         conv2d_1200[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1200 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1200[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1201 (Conv2D)            (None, 8, 8, 224)    129024      activation_1200[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1201 (Batch (None, 8, 8, 224)    672         conv2d_1201[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1201 (Activation)    (None, 8, 8, 224)    0           batch_normalization_1201[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1199 (Conv2D)            (None, 8, 8, 192)    399360      block8_5_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1202 (Conv2D)            (None, 8, 8, 256)    172032      activation_1201[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1199 (Batch (None, 8, 8, 192)    576         conv2d_1199[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1202 (Batch (None, 8, 8, 256)    768         conv2d_1202[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1199 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1199[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1202 (Activation)    (None, 8, 8, 256)    0           batch_normalization_1202[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block8_6_mixed (Concatenate)    (None, 8, 8, 448)    0           activation_1199[0][0]            \n",
      "                                                                 activation_1202[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_6_conv (Conv2D)          (None, 8, 8, 2080)   933920      block8_6_mixed[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block8_6 (Lambda)               (None, 8, 8, 2080)   0           block8_5_ac[0][0]                \n",
      "                                                                 block8_6_conv[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "block8_6_ac (Activation)        (None, 8, 8, 2080)   0           block8_6[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1204 (Conv2D)            (None, 8, 8, 192)    399360      block8_6_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1204 (Batch (None, 8, 8, 192)    576         conv2d_1204[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1204 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1204[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1205 (Conv2D)            (None, 8, 8, 224)    129024      activation_1204[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1205 (Batch (None, 8, 8, 224)    672         conv2d_1205[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1205 (Activation)    (None, 8, 8, 224)    0           batch_normalization_1205[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1203 (Conv2D)            (None, 8, 8, 192)    399360      block8_6_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1206 (Conv2D)            (None, 8, 8, 256)    172032      activation_1205[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1203 (Batch (None, 8, 8, 192)    576         conv2d_1203[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1206 (Batch (None, 8, 8, 256)    768         conv2d_1206[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1203 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1203[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1206 (Activation)    (None, 8, 8, 256)    0           batch_normalization_1206[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block8_7_mixed (Concatenate)    (None, 8, 8, 448)    0           activation_1203[0][0]            \n",
      "                                                                 activation_1206[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_7_conv (Conv2D)          (None, 8, 8, 2080)   933920      block8_7_mixed[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block8_7 (Lambda)               (None, 8, 8, 2080)   0           block8_6_ac[0][0]                \n",
      "                                                                 block8_7_conv[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "block8_7_ac (Activation)        (None, 8, 8, 2080)   0           block8_7[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1208 (Conv2D)            (None, 8, 8, 192)    399360      block8_7_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1208 (Batch (None, 8, 8, 192)    576         conv2d_1208[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1208 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1208[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1209 (Conv2D)            (None, 8, 8, 224)    129024      activation_1208[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1209 (Batch (None, 8, 8, 224)    672         conv2d_1209[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1209 (Activation)    (None, 8, 8, 224)    0           batch_normalization_1209[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1207 (Conv2D)            (None, 8, 8, 192)    399360      block8_7_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1210 (Conv2D)            (None, 8, 8, 256)    172032      activation_1209[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1207 (Batch (None, 8, 8, 192)    576         conv2d_1207[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1210 (Batch (None, 8, 8, 256)    768         conv2d_1210[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1207 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1207[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1210 (Activation)    (None, 8, 8, 256)    0           batch_normalization_1210[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block8_8_mixed (Concatenate)    (None, 8, 8, 448)    0           activation_1207[0][0]            \n",
      "                                                                 activation_1210[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_8_conv (Conv2D)          (None, 8, 8, 2080)   933920      block8_8_mixed[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block8_8 (Lambda)               (None, 8, 8, 2080)   0           block8_7_ac[0][0]                \n",
      "                                                                 block8_8_conv[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "block8_8_ac (Activation)        (None, 8, 8, 2080)   0           block8_8[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1212 (Conv2D)            (None, 8, 8, 192)    399360      block8_8_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1212 (Batch (None, 8, 8, 192)    576         conv2d_1212[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1212 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1212[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1213 (Conv2D)            (None, 8, 8, 224)    129024      activation_1212[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1213 (Batch (None, 8, 8, 224)    672         conv2d_1213[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1213 (Activation)    (None, 8, 8, 224)    0           batch_normalization_1213[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1211 (Conv2D)            (None, 8, 8, 192)    399360      block8_8_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1214 (Conv2D)            (None, 8, 8, 256)    172032      activation_1213[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1211 (Batch (None, 8, 8, 192)    576         conv2d_1211[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1214 (Batch (None, 8, 8, 256)    768         conv2d_1214[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1211 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1211[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1214 (Activation)    (None, 8, 8, 256)    0           batch_normalization_1214[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block8_9_mixed (Concatenate)    (None, 8, 8, 448)    0           activation_1211[0][0]            \n",
      "                                                                 activation_1214[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_9_conv (Conv2D)          (None, 8, 8, 2080)   933920      block8_9_mixed[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block8_9 (Lambda)               (None, 8, 8, 2080)   0           block8_8_ac[0][0]                \n",
      "                                                                 block8_9_conv[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "block8_9_ac (Activation)        (None, 8, 8, 2080)   0           block8_9[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1216 (Conv2D)            (None, 8, 8, 192)    399360      block8_9_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1216 (Batch (None, 8, 8, 192)    576         conv2d_1216[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1216 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1216[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1217 (Conv2D)            (None, 8, 8, 224)    129024      activation_1216[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1217 (Batch (None, 8, 8, 224)    672         conv2d_1217[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1217 (Activation)    (None, 8, 8, 224)    0           batch_normalization_1217[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1215 (Conv2D)            (None, 8, 8, 192)    399360      block8_9_ac[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1218 (Conv2D)            (None, 8, 8, 256)    172032      activation_1217[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1215 (Batch (None, 8, 8, 192)    576         conv2d_1215[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1218 (Batch (None, 8, 8, 256)    768         conv2d_1218[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1215 (Activation)    (None, 8, 8, 192)    0           batch_normalization_1215[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "activation_1218 (Activation)    (None, 8, 8, 256)    0           batch_normalization_1218[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block8_10_mixed (Concatenate)   (None, 8, 8, 448)    0           activation_1215[0][0]            \n",
      "                                                                 activation_1218[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_10_conv (Conv2D)         (None, 8, 8, 2080)   933920      block8_10_mixed[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_10 (Lambda)              (None, 8, 8, 2080)   0           block8_9_ac[0][0]                \n",
      "                                                                 block8_10_conv[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "conv_7b (Conv2D)                (None, 8, 8, 1536)   3194880     block8_10[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv_7b_bn (BatchNormalization) (None, 8, 8, 1536)   4608        conv_7b[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv_7b_ac (Activation)         (None, 8, 8, 1536)   0           conv_7b_bn[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "global_average_pooling2d_1 (Glo (None, 1536)         0           conv_7b_ac[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dense_16 (Dense)                (None, 5)            7685        global_average_pooling2d_1[0][0] \n",
      "==================================================================================================\n",
      "Total params: 54,344,421\n",
      "Trainable params: 54,283,877\n",
      "Non-trainable params: 60,544\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Finetuning on 5-class dataset\n",
    "base_model = InceptionResNetV2(include_top=False, pooling='avg', weights='imagenet', input_shape=(299, 299, 3))\n",
    "outputs = Dense(5, activation='softmax')(base_model.output)\n",
    "model = keras.Model(base_model.inputs, outputs)\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "DU8_tQHWsd8d",
    "outputId": "b9906b7f-45df-4ac6-8643-4396fb4fb3cc"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<keras.engine.input_layer.InputLayer object at 0x7f44d7da9898> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d7da9080> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d7d92518> False\n",
      "<keras.layers.core.Activation object at 0x7f44d7d92470> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d4ce1198> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d4cc07b8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d4c452b0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f4825e88a20> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f4825e57358> False\n",
      "<keras.layers.core.Activation object at 0x7f4825e76208> False\n",
      "<keras.layers.pooling.MaxPooling2D object at 0x7f4825de5d30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f4825dccd68> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f4825d2a940> False\n",
      "<keras.layers.core.Activation object at 0x7f4825d2ada0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f4825cc59e8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f4825cbddd8> False\n",
      "<keras.layers.core.Activation object at 0x7f4825c919b0> False\n",
      "<keras.layers.pooling.MaxPooling2D object at 0x7f4825c07d30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f4825904b00> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f4825967f98> False\n",
      "<keras.layers.core.Activation object at 0x7f48258f1d68> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f4825b05c50> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f48258bacc0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f4825b1bb38> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f4825809668> False\n",
      "<keras.layers.core.Activation object at 0x7f4825ad8f60> False\n",
      "<keras.layers.core.Activation object at 0x7f482581dfd0> False\n",
      "<keras.layers.pooling.AveragePooling2D object at 0x7f44e7a4bf98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f4825c07be0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f4825a22f98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f4825780b70> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e7a67b00> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f4825b707b8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f4825a02cc0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f4825772438> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e79d3da0> False\n",
      "<keras.layers.core.Activation object at 0x7f4825b70ac8> False\n",
      "<keras.layers.core.Activation object at 0x7f4825a39a90> False\n",
      "<keras.layers.core.Activation object at 0x7f482577dda0> False\n",
      "<keras.layers.core.Activation object at 0x7f44e79d3358> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44e79af048> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e76b8f98> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e769bcc0> False\n",
      "<keras.layers.core.Activation object at 0x7f44e7622f98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e78944a8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e75a4908> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e7823358> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e7580fd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44e7804240> False\n",
      "<keras.layers.core.Activation object at 0x7f44e75408d0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e79afda0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e774af60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e74d3cc0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e793e1d0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e77b0ef0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e74a5668> False\n",
      "<keras.layers.core.Activation object at 0x7f44e791d198> False\n",
      "<keras.layers.core.Activation object at 0x7f44e776dd68> False\n",
      "<keras.layers.core.Activation object at 0x7f44e74b9fd0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44e741db70> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e741d630> False\n",
      "<keras.layers.core.Lambda object at 0x7f44e7369ef0> False\n",
      "<keras.layers.core.Activation object at 0x7f44e73a1978> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e68c0860> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e691ef60> False\n",
      "<keras.layers.core.Activation object at 0x7f44e68abb38> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e72ede10> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e6854c18> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e72cfe48> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e67c2470> False\n",
      "<keras.layers.core.Activation object at 0x7f44e6a56eb8> False\n",
      "<keras.layers.core.Activation object at 0x7f44e67d7dd8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e73a1da0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e69d6f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e6794f60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e7322208> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e6a37ef0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e672f390> False\n",
      "<keras.layers.core.Activation object at 0x7f44e7322fd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44e69f5898> False\n",
      "<keras.layers.core.Activation object at 0x7f44e6710278> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44e6689f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e6689a20> False\n",
      "<keras.layers.core.Lambda object at 0x7f44e6627c88> False\n",
      "<keras.layers.core.Activation object at 0x7f44e65fdeb8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e63f9b00> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e6365320> False\n",
      "<keras.layers.core.Activation object at 0x7f44e63758d0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e6578cf8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e62c1e48> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e6555f28> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e62f6ba8> False\n",
      "<keras.layers.core.Activation object at 0x7f44e64e4cc0> False\n",
      "<keras.layers.core.Activation object at 0x7f44e62b5cf8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e65fdcc0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e64abc50> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e61fce48> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e658c8d0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e63fd5f8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e61dee80> False\n",
      "<keras.layers.core.Activation object at 0x7f44e658c780> False\n",
      "<keras.layers.core.Activation object at 0x7f44e6412f60> False\n",
      "<keras.layers.core.Activation object at 0x7f44e619ec50> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44e60e7f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e6116b00> False\n",
      "<keras.layers.core.Lambda object at 0x7f44e6054f28> False\n",
      "<keras.layers.core.Activation object at 0x7f44e6084278> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e5e62e48> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e5e18b00> False\n",
      "<keras.layers.core.Activation object at 0x7f44e5dd6fd0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e5fccba8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e5d21f60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e5f9e550> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e5d02c88> False\n",
      "<keras.layers.core.Activation object at 0x7f44e5fb3eb8> False\n",
      "<keras.layers.core.Activation object at 0x7f44e5cbfa58> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e6084a20> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e5f6dfd0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e5c07ac8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e601ab38> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e5e3ffd0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44e5c68f60> False\n",
      "<keras.layers.core.Activation object at 0x7f44e601a8d0> False\n",
      "<keras.layers.core.Activation object at 0x7f44e5eebb00> False\n",
      "<keras.layers.core.Activation object at 0x7f44e5c29828> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44e5bbad30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e5ba2e48> False\n",
      "<keras.layers.core.Lambda object at 0x7f44e5adfac8> False\n",
      "<keras.layers.core.Activation object at 0x7f44e5b0ee10> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d6afceb8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d6b5e438> False\n",
      "<keras.layers.core.Activation object at 0x7f44d6ae9eb8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d6d748d0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d6a68828> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d6cc5198> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d6a46ef0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d6cc5fd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d6a057f0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44e5b0e7b8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d6c0edd8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d699bbe0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d6d5dcf8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d6c73e10> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d696a588> False\n",
      "<keras.layers.core.Activation object at 0x7f44d6d5d4a8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d6c32be0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d68b9f98> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d68e5a90> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d68e5c50> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d687f940> False\n",
      "<keras.layers.core.Activation object at 0x7f44d682ddd8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d65cec18> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d65e4e80> False\n",
      "<keras.layers.core.Activation object at 0x7f44d6574a58> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d67b0d30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d651cb38> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d6792d68> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d648f2e8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d671edd8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d64a2898> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d6864860> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d669fe80> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d6468ef0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d67e4080> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d667dfd0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d6387dd8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d67e4ef0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d663ba20> False\n",
      "<keras.layers.core.Activation object at 0x7f44d63d99b0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d6351d30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d6351f28> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d62f1ba8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d62c2f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d603ea20> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d6010358> False\n",
      "<keras.layers.core.Activation object at 0x7f44d602e208> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d61bf908> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d5f8bd68> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d621cfd0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d5f3fac8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d61a8be0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d5efdf98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d62c2f60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d6172b70> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d5e41d68> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d62544e0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d60c4518> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d5ea4da0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d6254908> False\n",
      "<keras.layers.core.Activation object at 0x7f44d6094cc0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d5e62b70> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d5dadfd0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d5ddff28> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d4a55e48> False\n",
      "<keras.layers.core.Activation object at 0x7f44d4a86ba8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d47b7f98> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d481df28> False\n",
      "<keras.layers.core.Activation object at 0x7f44d47d9f60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d4a30ba8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d4723e80> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d49a0470> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d4708f60> False\n",
      "<keras.layers.core.Activation object at 0x7f44d49afdd8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d46c1b70> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d4a86b38> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d4973eb8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d460f940> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d4a1db38> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d48eb160> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d466af98> False\n",
      "<keras.layers.core.Activation object at 0x7f44d4a1d7f0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d48ebda0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d462a908> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d44f4710> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d453dc50> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d4554320> False\n",
      "<keras.layers.core.Activation object at 0x7f44d44e09e8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d4242f60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d42a6f28> False\n",
      "<keras.layers.core.Activation object at 0x7f44d4230dd8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d4441cc0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d41ac748> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d4452ba8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d418ce48> False\n",
      "<keras.layers.core.Activation object at 0x7f44d4410cf8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d414b828> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d450f7f0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d435dda0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d40c4be0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d44a2d68> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d433dd30> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d40b24a8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d44a2048> False\n",
      "<keras.layers.core.Activation object at 0x7f44d42f9b00> False\n",
      "<keras.layers.core.Activation object at 0x7f44d4047e10> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d4003f98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d4029b38> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d3ef8cf8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3f7f828> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3d19c88> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d3cea630> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3c3d978> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3e7ef60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3c67b70> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d3eddc88> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d3bd8400> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3e69f60> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3bb92e8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3f7f2b0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3de78d0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3bb2f60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d3f12b00> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d3dc9f60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d3b65be0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3f2ef98> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3d86898> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3b1ed30> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d3a67f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3a9ac50> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d39d9f28> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3a08358> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3903828> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d3964ef0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d38f2b00> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3898be0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d380c438> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3818da0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3a08c88> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d37e5f98> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d399ce48> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d3770358> False\n",
      "<keras.layers.core.Activation object at 0x7f44d399cb38> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3755240> False\n",
      "<keras.layers.pooling.MaxPooling2D object at 0x7f44d36cce80> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d36ccc88> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d35c9a90> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d356ffd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d359e780> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d34e7f28> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d34cdeb8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d348ac88> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d367cb38> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d33d4f60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d363ac88> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d33b6c88> False\n",
      "<keras.layers.core.Activation object at 0x7f44d362eba8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d33f1a58> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d32b7ac8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d336af98> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d32a8cf8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d32d9240> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3204a58> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d3177278> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3186828> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3151da0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d3106b00> False\n",
      "<keras.layers.core.Activation object at 0x7f44d30c2fd0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d32d9828> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d3005da0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d326eef0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d306ddd8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d326e6a0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d3028ba8> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d2f72e80> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2fa6f60> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d2edde80> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2f0ebe0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2e3cba8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d2e2d470> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2dbedd8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2d79eb8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d2cf4160> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2cf4da0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2f0eb70> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2c3ff98> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d2eaab38> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d2ca5f28> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2eaa7f0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2c60f60> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d2badfd0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2be0ba8> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d2b16e10> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2b46c50> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2a47710> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d2aa6dd8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2a33a90> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d29dab70> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d294f2e8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d295e978> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2b46cf8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2918f60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d2adcc50> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d28932b0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2adc9e8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d28b12b0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d2828d30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d280eac8> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d27aedd8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d277af60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2679fd0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d26dcf98> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2665ef0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d25e2860> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d25c1f60> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2580908> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d277aa90> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2513cc0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d27106d8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d247ac18> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2710da0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d24e57f0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d245fc18> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d245f6d8> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d239afd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d23e1ac8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d22fcb70> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d230ea58> False\n",
      "<keras.layers.core.Activation object at 0x7f44d22caf98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2217f98> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d21f7cc0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d21b4cc0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d23e1ef0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d20faf98> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d2360320> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d215df60> False\n",
      "<keras.layers.core.Activation object at 0x7f44d23606a0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2117b38> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d2062978> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2093fd0> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d1fd1cf8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d2001a58> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1fa3a58> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d1f1d278> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1f2e908> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1e74e80> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d1eabbe0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1e62a58> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d2001780> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1ddecf8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d1f99ef0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d1d95a58> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1f996a0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1d50f60> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d1c90da0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1cc2c88> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d1c01ef0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1c36d30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1b7bc50> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d1b8fe80> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1b1ab70> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1ac8c50> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d1a344e0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1a02cf8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1c36dd8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1a2da58> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d1bc4d30> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d199f240> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1bc4ac8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d19348d0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d18f8f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d18f8e48> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d1861cc0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d18e7fd0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d17e6f28> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d17c7c18> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1749fd0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d16cc940> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d172cf98> False\n",
      "<keras.layers.core.Activation object at 0x7f44d16eb7f0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d18e71d0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d15fec50> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d17fa860> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d1612fd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d17fae80> False\n",
      "<keras.layers.core.Activation object at 0x7f44d15cf860> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d1567c18> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1548cf8> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d14869b0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d14cbba8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1468c50> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d13fab38> False\n",
      "<keras.layers.core.Activation object at 0x7f44d13b89e8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d12fdd68> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d1361da0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d131cda0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d14cbfd0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1266ef0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d144f400> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d124afd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d144f780> False\n",
      "<keras.layers.core.Activation object at 0x7f44d12049e8> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d114cb00> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d117ee48> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d10b6e10> False\n",
      "<keras.layers.core.Activation object at 0x7f44d116c9e8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d1098b38> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d100d470> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1017e48> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0fd0f28> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d0f511d0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0f51f98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d116ca90> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0ecbe80> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d1084fd0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d0e7cb38> False\n",
      "<keras.layers.core.Activation object at 0x7f44d1084780> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0e399b0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d0d7ef28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0e30c50> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d0d4eb70> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0da1c88> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0c9c8d0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d0c7af60> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0c07c50> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0bd0be0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d0ba15c0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0b36f98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0da1d30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0b1db38> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d0d2feb8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d0a8c320> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0d2fba8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0a9b9b0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d0a58f98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0a04b38> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d09d5ac8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d094cda0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d08d3da0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d092fcf8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d083ae48> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d07b6ac8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d0818f60> False\n",
      "<keras.layers.core.Activation object at 0x7f44d07d7828> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0982828> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d071f898> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d0965b70> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d06fff28> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0904390> False\n",
      "<keras.layers.core.Activation object at 0x7f44d06bf940> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d0605710> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0650c50> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d05f3a90> False\n",
      "<keras.layers.core.Activation object at 0x7f44d06227b8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0553d30> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d0520160> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0520940> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0468e48> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d044ae80> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0409e80> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d06227f0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0353fd0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d0536dd8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d03b2cf8> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0536048> False\n",
      "<keras.layers.core.Activation object at 0x7f44d036fcf8> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44d0239fd0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d02e8f28> False\n",
      "<keras.layers.core.Lambda object at 0x7f44d0223ef0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0254198> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0181c18> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d016e550> False\n",
      "<keras.layers.core.Activation object at 0x7f44d0103f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d00bf908> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d003e2b0> False\n",
      "<keras.layers.core.Activation object at 0x7f44d005c2b0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44d0254940> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cffced30> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44d01ebac8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cffa0128> False\n",
      "<keras.layers.core.Activation object at 0x7f44d01eb860> False\n",
      "<keras.layers.core.Activation object at 0x7f44cffa0a90> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cff19d30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cff19f28> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cfe38c50> False\n",
      "<keras.layers.core.Activation object at 0x7f44cfe8aef0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cfd88e80> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cfde5fd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cfd6fd68> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cfcb8cc0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cfca0c18> False\n",
      "<keras.layers.core.Activation object at 0x7f44cfc8c7f0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cfe8af98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cfc07c18> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cfe19668> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cfbf2518> False\n",
      "<keras.layers.core.Activation object at 0x7f44cfe19710> False\n",
      "<keras.layers.core.Activation object at 0x7f44cfb42f98> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cfb6ea90> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cfb6ec50> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cfabfba8> False\n",
      "<keras.layers.core.Activation object at 0x7f44cfab3e80> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf9b5dd8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cfa1ddd8> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf9a7f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf918f98> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cf906f60> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf8c0b38> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cfaec908> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf80d978> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cfa6f198> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cf869f98> False\n",
      "<keras.layers.core.Activation object at 0x7f44cfa6ff60> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf827780> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cf73cd30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf7a0e48> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cf6ddb70> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf688f60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf638550> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cf60c240> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf62c240> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf583eb8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cf539a58> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf4f5f60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf70df28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf438da0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cf6a5eb8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cf4a1dd8> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf6a5518> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf45cdd8> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cf3a4f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf3d8b00> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cf310fd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf341278> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf28bc18> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cf25a630> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf2289e8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf1d2ac8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cf147240> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf1588d0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf341a20> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf11ff28> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cf2d9ba8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cf0b02e8> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf2d9940> False\n",
      "<keras.layers.core.Activation object at 0x7f44cf08e1d0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cf005f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cf005a20> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cefa4d30> False\n",
      "<keras.layers.core.Activation object at 0x7f44cef76eb8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ceeeff28> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ceed4ef0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cee5ae48> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cee23c50> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cedb7fd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44ced77860> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cef769e8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ced0bc18> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cef06630> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cecdc5f8> False\n",
      "<keras.layers.core.Activation object at 0x7f44cef06f98> False\n",
      "<keras.layers.core.Activation object at 0x7f44cecf1fd0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cec58b70> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cec58630> False\n",
      "<keras.layers.core.Lambda object at 0x7f44ceb96f28> False\n",
      "<keras.layers.core.Activation object at 0x7f44cebdba20> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce8518d0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce835f60> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce7bfc50> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cebdbe48> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cea04d30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce785be0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ceb59278> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cea6ad68> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce7565c0> False\n",
      "<keras.layers.core.Activation object at 0x7f44ceb595f8> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce975e80> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce76cf98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ceb21f60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce96db00> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce6d0b38> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ceb06eb8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce951f98> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce63f320> False\n",
      "<keras.layers.core.Activation object at 0x7f44ceac1ef0> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce90da90> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce64e9b0> False\n",
      "<keras.layers.pooling.MaxPooling2D object at 0x7f44ce608f98> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44ce5b42e8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce532ba8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce484588> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce49af60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce454f60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce3ee198> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce37e828> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce5a1198> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce361d68> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce4fada0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce2b4160> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce4fa668> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce2b4940> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44ce2b1f60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce2b1d68> False\n",
      "<keras.layers.core.Lambda object at 0x7f44ce1cec88> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce221fd0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce11dcf8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce0faf28> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce086da0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce005710> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce063dd8> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce01f7f0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ce221f28> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cdf36b70> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ce1ae898> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cdf09550> False\n",
      "<keras.layers.core.Activation object at 0x7f44ce1ae748> False\n",
      "<keras.layers.core.Activation object at 0x7f44cded6fd0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cde81c88> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cde81ac8> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cde51be0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cddcaef0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cdd49e10> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cddaee10> False\n",
      "<keras.layers.core.Activation object at 0x7f44cdcbef60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cdc37fd0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cdc9cf98> False\n",
      "<keras.layers.core.Activation object at 0x7f44cdc54940> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cde03940> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cdba0e80> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cdd831d0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cdb7ffd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cdd83f98> False\n",
      "<keras.layers.core.Activation object at 0x7f44cdb3c7b8> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cda84828> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cdab6ef0> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cda71ba8> False\n",
      "<keras.layers.core.Activation object at 0x7f44cda19f60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd9cca90> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd99f278> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd93b278> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd918d30> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd8cea90> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd88df98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cdaa3ef0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd7d1dd8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd9b5748> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd7b6e10> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd9b5550> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd7f1e10> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cd6baf60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd76bb38> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cd6a1fd0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd6d7be0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd61cc50> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd584be0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd53ca20> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd568b00> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd4db278> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd4ea908> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd6d7c88> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd4b2f60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd66bbe0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd421240> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd66b978> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd3c3240> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cd39ba58> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd39bef0> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cd2b9d68> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd30fef0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd205f60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd268f28> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd1f1e80> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd13ad30> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd14ee80> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd10d898> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd30fa20> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cd0a2c50> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd29e668> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cd072630> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd29ef60> False\n",
      "<keras.layers.core.Activation object at 0x7f44cd005fd0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44ccfeb668> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ccfebba8> False\n",
      "<keras.layers.core.Lambda object at 0x7f44ccf26da0> False\n",
      "<keras.layers.core.Activation object at 0x7f44ccf6ea58> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cce34f98> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cce9bef0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cce57eb8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ccda3f28> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ccd88c50> False\n",
      "<keras.layers.core.Activation object at 0x7f44ccd42a20> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ccf6ee80> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ccc87f28> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cceed2b0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ccce9ef0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cceed630> False\n",
      "<keras.layers.core.Activation object at 0x7f44ccca9ac8> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44ccbf1908> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ccc20f60> False\n",
      "<keras.layers.core.Lambda object at 0x7f44ccb5bc88> False\n",
      "<keras.layers.core.Activation object at 0x7f44ccb8df60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ccab8b70> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ccaa8208> False\n",
      "<keras.layers.core.Activation object at 0x7f44cca3c898> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cca01e10> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cc9b7b70> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc9769e8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44ccb8d898> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc8baf60> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44ccb23e80> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cc91def0> False\n",
      "<keras.layers.core.Activation object at 0x7f44ccb23630> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc8dcef0> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cc820d30> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc85bc18> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cc792e80> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc7bfcc0> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc6ba780> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cc71be80> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc6abb00> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc651be0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cc5c7470> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc592f98> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc7bfd68> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc538b70> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cc753cc0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cc52c1d0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc753a58> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc4bb860> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cc486dd8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc486eb8> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cc4131d0> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc472198> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc370eb8> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cc351400> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc2ddf60> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc25c8d0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cc23cf60> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc1f9860> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc472978> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc140748> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cc3867f0> False\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cc1a1e48> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc386e10> False\n",
      "<keras.layers.core.Activation object at 0x7f44cc15e828> False\n",
      "<keras.layers.merge.Concatenate object at 0x7f44cc0f2ba8> False\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc0d6c88> False\n",
      "<keras.layers.core.Lambda object at 0x7f44cc014940> True\n",
      "<keras.layers.convolutional.Conv2D object at 0x7f44cc05ab38> True\n",
      "<keras.layers.normalization.BatchNormalization object at 0x7f44cbfc0d68> True\n",
      "<keras.layers.core.Activation object at 0x7f44cbfda2e8> True\n",
      "<keras.layers.pooling.GlobalAveragePooling2D object at 0x7f44cbfa1f28> True\n",
      "<keras.layers.core.Dense object at 0x7f44d7da9860> True\n"
     ]
    }
   ],
   "source": [
    "for i, layer in enumerate(model.layers):\n",
    "  if i < 776:\n",
    "    layer.trainable = False\n",
    "  else:\n",
    "    layer.trainable = True\n",
    "\n",
    "# check the trainable status of the individual layers\n",
    "for layer in model.layers:\n",
    "    print(layer, layer.trainable)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "TdyaTnwGS59C"
   },
   "source": [
    "# **Optimization & Model Compile**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "1RLZfD3F6_f1"
   },
   "outputs": [],
   "source": [
    "from keras.callbacks import LearningRateScheduler\n",
    "from math import exp\n",
    "epoch =20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "7UFeCaLO6k5p"
   },
   "outputs": [],
   "source": [
    "momentum = 0.9\n",
    "sgd = keras.optimizers.SGD(lr=0.001, momentum=momentum, decay=0.0, nesterov=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "DPT7M2SqOKD_"
   },
   "outputs": [],
   "source": [
    "# Regular Model\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer=sgd,\n",
    "              metrics=['acc'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "H0-9E-xGTLRB"
   },
   "source": [
    "# **Fit Model**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "MIkjaWDR2spe"
   },
   "outputs": [],
   "source": [
    "label_count=0\n",
    "total_count=0\n",
    "class_weights=0\n",
    "from collections import Counter\n",
    "labels_count = Counter(lbl for lbl in df_train['diagnosis'])\n",
    "total_count = sum(labels_count.values())\n",
    "class_weights = {cls: total_count / count for cls, count in labels_count.items()}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 837
    },
    "colab_type": "code",
    "id": "iR7pmgqXOUjH",
    "outputId": "4061b06d-3b2a-4aea-df82-6ee5c15bd977"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "27/26 [===============================] - 512s 19s/step - loss: 1.4489 - acc: 0.3726 - val_loss: 1.6182 - val_acc: 0.4925\n",
      "Epoch 2/20\n",
      "27/26 [===============================] - 454s 17s/step - loss: 1.2705 - acc: 0.5018 - val_loss: 1.6375 - val_acc: 0.4939\n",
      "Epoch 3/20\n",
      "27/26 [===============================] - 455s 17s/step - loss: 1.2195 - acc: 0.5147 - val_loss: 1.6715 - val_acc: 0.4966\n",
      "Epoch 4/20\n",
      "27/26 [===============================] - 443s 16s/step - loss: 1.1798 - acc: 0.5395 - val_loss: 1.7313 - val_acc: 0.5048\n",
      "Epoch 5/20\n",
      "27/26 [===============================] - 443s 16s/step - loss: 1.1368 - acc: 0.5514 - val_loss: 1.8377 - val_acc: 0.4980\n",
      "Epoch 6/20\n",
      "27/26 [===============================] - 456s 17s/step - loss: 1.1240 - acc: 0.5480 - val_loss: 1.6990 - val_acc: 0.5089\n",
      "Epoch 7/20\n",
      "27/26 [===============================] - 466s 17s/step - loss: 1.1020 - acc: 0.5724 - val_loss: 1.7227 - val_acc: 0.5102\n",
      "Epoch 8/20\n",
      "27/26 [===============================] - 466s 17s/step - loss: 1.1125 - acc: 0.5552 - val_loss: 1.6884 - val_acc: 0.5143\n",
      "Epoch 9/20\n",
      "27/26 [===============================] - 473s 18s/step - loss: 1.0847 - acc: 0.5733 - val_loss: 1.7903 - val_acc: 0.5075\n",
      "Epoch 10/20\n",
      "27/26 [===============================] - 465s 17s/step - loss: 1.0725 - acc: 0.5675 - val_loss: 1.7348 - val_acc: 0.5130\n",
      "Epoch 11/20\n",
      "27/26 [===============================] - 449s 17s/step - loss: 1.0472 - acc: 0.5736 - val_loss: 1.7594 - val_acc: 0.5130\n",
      "Epoch 12/20\n",
      "27/26 [===============================] - 445s 16s/step - loss: 1.0746 - acc: 0.5846 - val_loss: 1.7035 - val_acc: 0.5211\n",
      "Epoch 13/20\n",
      "27/26 [===============================] - 439s 16s/step - loss: 1.0234 - acc: 0.6033 - val_loss: 1.6856 - val_acc: 0.5211\n",
      "Epoch 14/20\n",
      "27/26 [===============================] - 454s 17s/step - loss: 1.0652 - acc: 0.5736 - val_loss: 1.5719 - val_acc: 0.5307\n",
      "Epoch 15/20\n",
      "27/26 [===============================] - 444s 16s/step - loss: 1.0380 - acc: 0.5905 - val_loss: 1.6317 - val_acc: 0.5307\n",
      "Epoch 16/20\n",
      "27/26 [===============================] - 452s 17s/step - loss: 1.0266 - acc: 0.5879 - val_loss: 1.6472 - val_acc: 0.5280\n",
      "Epoch 17/20\n",
      "27/26 [===============================] - 463s 17s/step - loss: 1.0383 - acc: 0.5756 - val_loss: 1.5853 - val_acc: 0.5389\n",
      "Epoch 18/20\n",
      "27/26 [===============================] - 449s 17s/step - loss: 1.0165 - acc: 0.5996 - val_loss: 1.6081 - val_acc: 0.5321\n",
      "Epoch 19/20\n",
      "27/26 [===============================] - 446s 17s/step - loss: 1.0084 - acc: 0.6024 - val_loss: 1.6779 - val_acc: 0.5293\n",
      "Epoch 20/20\n",
      "27/26 [===============================] - 451s 17s/step - loss: 0.9952 - acc: 0.6047 - val_loss: 1.6442 - val_acc: 0.5334\n"
     ]
    }
   ],
   "source": [
    "history = model.fit_generator(\n",
    "      train_generator,\n",
    "      steps_per_epoch=train_generator.samples/train_generator.batch_size ,\n",
    "      epochs=epoch,\n",
    "      validation_data=validation_generator,\n",
    "      validation_steps=validation_generator.samples/validation_generator.batch_size,\n",
    "      verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "SYIiq_9qHoSG"
   },
   "source": [
    "# **Save Weights For Future Use**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "0VCzKXDNHm0P"
   },
   "outputs": [],
   "source": [
    "# Save Weights of the Model for Future Use\n",
    "model.save_weights('Model_InceptionResnetV2.h5') # V2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "rXyxrc4gTUJR"
   },
   "source": [
    "# **Model Accuracy and Loss Plot**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 585
    },
    "colab_type": "code",
    "id": "ua3-CUVZAnXQ",
    "outputId": "a8766082-11cf-4d7f-da31-2a122f4cf696"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZYAAAEcCAYAAADk05IoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzs3Xd4VGXa+PHvlPRC2qQSSOgBQkgI\nAQwoUgQETCyIYsOXoqCyuhbw1RVXeVfZ39pWRXddRbGsioUSQEDEpQlIkRY6CQmQZCZlUieZdn5/\nBGaJtEmfwP25Lq4kZ55zzj2TMPc8XaUoioIQQgjRRNStHYAQQoiriyQWIYQQTUoSixBCiCYliUUI\nIUSTksQihBCiSUliEUII0aQksYgmY7PZSExM5MyZM01atjWdPHmS7t27N/l1t2zZwrBhwxw/jxo1\nih07djhVtr6ee+453n///QafL0R9aVs7ANF6EhMTHd+bTCbc3d3RaDQA/PnPf+aWW26p1/U0Gg27\nd+9u8rLXgtWrVzfJdRYvXsyyZcv49NNPHcf+7//+r0muLYSzJLFcw85/Yx82bBjz5s3juuuuu2R5\nq9WKVit/MsI1yN+j65KmMHFJb7zxBo8//jh//OMfSUxMZNmyZezevZs777yT5ORkBg8ezLx587BY\nLEDtf/Tu3btz6tQpAJ566inmzZvH1KlTSUxMZOLEieTm5ta7LMB//vMfRo0aRb9+/Xj55Ze56667\n+O677y4atzMxfvnll4wcOZL+/fszb948x7k2m42//OUvDBgwgOHDh7Nx48ZLvj7vvfceTzzxRJ1j\nf/7zn3nllVeA2trDmDFjSExMZMSIESxevPiS17r++uvZtm0bUFt7fPrpp+nfvz9jx45l//79dcou\nWLCA4cOHk5iYyNixY1m3bh0Ahw8f5qWXXmLHjh0kJiYyYMAAx2v79ttvO87/97//zciRIxkwYAAz\nZ85Er9c79drU53U+F8/kyZNJSUkhNTWVDz74wHGfd999lxEjRpCUlMRtt91GQUHBRZsd7777bsfv\nefHixdxzzz3MmzePlJQU3nvvPbKzs7nvvvtISUlhwIABPP3005SXlzvOP336NDNnzmTgwIEMGDCA\nefPmUVNTQ3JyMsePH3eU0+v1JCQkUFJScsnnK+pBEUJRlBtvvFHZvHlznWOvv/660qtXL2XdunWK\nzWZTTCaTsmfPHuW3335TLBaLkpOTo9x0003Kp59+qiiKolgsFqVbt25Kbm6uoiiK8uSTTyopKSnK\n3r17FbPZrPzhD39QnnzyyXqXLSwsVPr27ausXbtWMZvNykcffaT07NlT+fbbby/6XJyJ8eGHH1bK\nysqU3NxcpX///o7n/umnnyo333yzkpeXpxQXFyuTJk1SunXrdtH75OTkKH379lUqKysd1x44cKCy\nd+9eRVEUZd26dUpOTo5it9uVLVu2KPHx8crBgwcVRVGUzZs3KzfeeKPjWkOGDFG2bt2qKIqivPrq\nq8q9996rGI1G5dSpU8qYMWPqlF2xYoVSUFCg2Gw2ZdmyZUrfvn0Vg8GgKIqifP3118q9995bJ84n\nn3xS+fvf/64oiqJs3LhRGThwoJKZmalUV1crc+fOVe677z6nXpv6vM5lZWXKoEGDlI8//lipqalR\nysvLlT179iiKoijvv/++Mn78eCUrK0ux2WxKZmamUlJSomRnZ1/wWt91112O3/PXX3+txMXFKZ9/\n/rlitVoVk8mknDhxQtmyZYtSU1OjFBYWKhMnTlReffVVx/MZO3as8uqrryqVlZWKyWRSduzYoSiK\nojz//PPK66+/7rjPhx9+qMycOfOiz1PUn9RYxGUlJSUxbNgw1Go1np6e9OnTh4SEBLRaLdHR0dx5\n551s3779kuePGjWK+Ph43NzcGD9+PIcOHap32fXr1xMXF8eIESNwc3Nj8uTJBAYGXvI6zsT40EMP\n4efnR/v27UlJSeHgwYMArFq1ismTJxMeHk5gYCDTp0+/5H2io6Pp2rWro8awefNm/P39iY+PB2qb\nF6Ojo1GpVAwaNIhBgwaxc+fOS17vnFWrVjFjxgzatWtHVFQU99xzT53Hb775ZkJDQ1Gr1YwfP56o\nqKgLajWXsnz5ciZMmEBcXBweHh489dRT/Prrr+Tn51/xtfm9y73O69atIyIiggceeAB3d3d8fX3p\n06cPUFvz+OMf/0hMTAxqtZq4uDgCAgKcij8iIoJJkyah0Wjw9PQkNjaWQYMG4e7uTnBwMJMnT3bE\n8Ntvv1FSUsJTTz2Ft7c3np6e9OvXD4D09HSWL1+OcnapxKVLl5KWluZUDOLKpIFSXFZERESdn48f\nP878+fM5cOAAJpMJm83meMO4GJ1O5/jey8uLqqqqepfV6/V14lCpVISFhV3yOs7EGBIS4vje09Oz\nzr3Cw8Mdj0VGRl7yPgDjxo1jxYoVjB8/noyMDMaPH+94bP369SxYsICTJ09it9uprq52JJ3LMRgM\ndWKIioqq8/h3333Hxx9/7BhRV1VV5XQTjl6vrzNow9fXF39/fwoKChyvyaVem9+73Oucn59Phw4d\nLnpefn4+0dHRTsX7e7//ezQYDMybN49du3ZRWVmJoiiODx15eXlERUU5BqScr1+/fmi1Wnbu3Im/\nvz95eXkMHTq0QTGJC0mNRVyWSqWq8/PcuXPp2rUra9asYdeuXcyaNavZY9DpdHU+USuKQkFBwSXL\nNybG39/rSsOhx4wZw5YtWygoKGDdunWMGzcOgOrqambNmsVDDz3E5s2b2bFjB6mpqY5PyJcTEhJy\nyRhyc3N58cUXefHFF9m2bRs7duygU6dOTl0XIDQ0tM71KioqKCsru2yivpTLvc7h4eHk5ORc9Lzw\n8PA6/WfneHl5AbV9TOcUFhbWKfP7v8e//e1vuLu7s3z5cnbt2uXo34LaJHTmzBlsNttF40hLS2PZ\nsmUsXbqU0aNH4+7ufoVnLJwliUXUS2VlJX5+fnh7e3P8+HG++uqrZr/njTfeSGZmJj/99BNWq5VP\nPvnksp/QGxPjmDFj+OSTTygoKKCkpIR//etfly2v0+no168fc+bMITY2lpiYGADMZjMWi4XAwEA0\nGg3r16/nl19+cTqG999/n7KyMs6cOcPnn39e57mpVCqCgoJQFIWvv/6aEydOOB4/l5TO70Q/39ix\nY/nmm284dOgQZrOZ119/neTk5Do1JGdd7nUePnw4eXl5fPbZZ5jNZioqKti7dy8AEyZM4M033yQn\nJwdFUTh48CBGoxGdTodOp2PZsmXYbDa++uqrKyb2yspKvLy88PPzIy8vj48++sjxWN++fQkICOD1\n11/HZDJRXV1dpykyLS2N1atXk5GRQXp6er2fv7g0SSyiXmbPns33339PUlISL7zwAmPGjGn2e4aE\nhPDGG2/w6quvMmDAAHJzc4mLi7vkJ8zGxHj33XczcOBAxo8fzx133MGoUaOueM64cePYsmVLnWYw\nf39/nn32WR599FFSUlJYvXq1000tjz32GDqdjmHDhjFt2rQ6bf89evTg3nvvZcKECQwePJisrKw6\nzXypqal07NiR1NRUUlNTL7j29ddfzyOPPMKjjz7K4MGDOXPmDH/729+ciuv3Lvc6+/n58dFHH7F6\n9Wquu+46Ro0a5ej7mDJlCiNGjGDy5MkkJSXxpz/9iZqaGlQqFS+//DL/+Mc/GDhwICdPnrxsM+u5\n12rfvn0kJyczY8YMbrrpJsdjWq2W999/n+PHjzN06FCGDh1aZ75Q+/bt6datG+7u7iQlJTXoNRAX\np1KcrUML4SJsNhtDhgzh73//O8nJya0djmjDnnnmGaKjo3nsscdaO5SritRYRJuwYcMGysrKMJvN\nLFiwAK1We8VPs0JcTm5uLuvWreP2229v7VCuOjIqTLQJO3fu5KmnnsJqtdK1a1feeecd6WwVDfba\na6/x2WefMWPGjCuO/BP1J01hQgghmpQ0hQkhhGhSkliEEEI0KUksQgghmtQ11XlfUlKJ3V7/LqXg\nYF+KiiqaIaKmIfE1jsTXeK4eo8TXMGq1isBAn3qfd00lFrtdaVBiOXeuK5P4GkfiazxXj1HiaznS\nFCaEEKJJSWIRQgjRpCSxCCGEaFKSWIQQQjQpSSxCCCGalCQWIYQQTarFhhtnZWUxZ84cjEYjAQEB\nzJ8/37Ep0vlWrlzJe++9h6IoqFQqFi5cSEhICDabjXnz5rFx40ZUKhXTp09nwoQJLRW+EEI0G7td\nobDUREGxiYKSqvO+VlFYWo2vlxshAZ7o2nmd99ULXTtPAv090Khdq47QYoll7ty5TJo0ibS0NJYu\nXcoLL7zAokWL6pTZt28f77zzDp988gk6nY7y8nLHCrbLly8nJyeHNWvWYDQaSU9PZ9CgQbRv376l\nnoIQQjSYoiiUVpopKK6ioMREQXEV+cVV6EtM6I0mLFa7o6y7m5qwQG+iQ31J7Kaj0mTBYDRx7HQp\n2w/qsZ+3drBGrSLI34OQdl7oAjzRBXgRcl4C8vN2u2BL5+bWIomlqKiIzMxMFi5cCNTuuPfyyy9T\nXFxMUFCQo9zHH3/M//zP/6DT6YDaXejOWblyJRMmTECtVhMUFMSIESP44YcfmDp1aks8BSGEqLfK\nagtf/3SMkwXlFJSYqDHbHI9p1CpCA70IC/QmpXcEfp4awgO9CQvyJsDX/ZLJwGqzU1xeQ6HRRGFp\nNQajCcPZ7387WkhZVd1tqb09tDx9dyIdw/0uer3m0CKJJS8vj7CwMDQaDQAajYbQ0FDy8vLqJJbj\nx4/Tvn177rnnHqqqqhg5ciQzZsxApVKRl5dXZ9+EiIgI8vPzWyJ8IYSot6pqC699+Ru5+griYgLp\n2j6A8CBvwgK9CAvyJtjfE7W6NnnodH4YDOVOXVerURMa4EVogNdFH682WyksrabQWJt0yk0WAvw8\nmux5ORVji97tCmw2G4cPH2bhwoWYzWamTp1KZGQk6enpTXL94GDfBp+r07Vctm8Iia9xJL7Gc/UY\nWzK+SpOFVz7fxSlDBf87OYWUXuFXPKcp44uOarJLNUiLJJaIiAgKCgqw2WxoNBpsNht6vZ6IiIg6\n5SIjIxk9ejTu7u64u7szfPhw9u7dS3p6OhEREZw5c8axHe3vazDOKCqqaNB6PPX5NNEaJL7Gkfga\nz9VjbMn4qqqtvP71b5zML2fmrb2JDfW54r1d9fVTq1UN+kDeIkMJgoODiYuLIyMjA4CMjAzi4uLq\nNINBbd/Lpk2bUBQFi8XC1q1b6dGjBwCjR49m8eLF2O12iouL+fHHHxk1alRLhC+EaCWZ2cU88sYG\n3ly8h52HDVht9iuf1IpMNVbeOJtUZqT3JrGrrrVDahUt1hT24osvMmfOHBYsWIC/vz/z588HYNq0\nacyaNYv4+HjGjh3L/v37ufnmm1Gr1QwePJg77rgDgLS0NPbs2cNNN90EwCOPPEJ0dHRLhS/ENSmn\noJy1v+Yy7roYwoK8W/Tep/QVvPv9Pnw83cgpKOfd7/fh7+1GanwEQxIiCW/heK6kNqnsITu/nIfT\nepPU7dpMKnCN7XkvTWGtQ+JrnNaIT1EUftp1mq9+OobVZifY34M59/QjuJ1ni8RYXFbN/326E0VR\neP7+ZNr5urPvRDEb95xhz7Ei7IpCt+gArk+IoF/3UDzcNJe9XnO/hqYaK28s3sOJ02U8nNaL5B6h\n9TrfVf8GG9oU5lKd90KI1ldhsrBw5UF2Hy2kT+dgRiZHs2DJfv725W7m3NuPdj7uzXr/qmorby7e\ng6nGypx7kgjyr01mfbuE0LdLCMaKGrbsz2fDnjP8K+Mgn689wsCe4VyfENmiQ2rPqTZbeetsUnmo\nAUnlaiSJRQjhcCTXyD+WHaCs0sxdw7owon80apWKJyYk8LevdvPal7t5ZlISvl5uzXJ/q83Ou9/v\nI6+oiscnJNAh7MJEEeDrwc0DOzJmQAeO5BrZsOcMm/blsX73aTqE+XJ9QiQDe4bh7dk8MZ6vxmzj\nzcV7OXa6jOm39KS/JBVA1goTQlC7pMjSTVnM/2IXblo1z93fj5tSOqA+O0mvS/t2PHZ7H/KLq3jj\n69raRFNTFIWPVx3i4MkSHhjdg16xQZctr1Kp6N4hkGnje/H6o6ncM7IbKPDZmiM88c5mPlieyeGc\nEpqrtb/GbOOtb/Zw9JSRaeN7khIX1iz3aYukxiLENa64rLr2TTjXyKBeYdx7U3e8PC58a+gVE8SM\n9N68+91+3v52L49PSMD9Cn0b9fH9xiy27M8nfXAsg/tEXPmE8/h4ujG8X3uG92vPyfxyNuw5w9bM\nfH45kE9YoBejr4ulb2wg7XybZqJgjaU2qRzONTJtXE8G9JSkcj5JLEJcw3YfNfDRioNYbQpTxsaR\nGn/5N/TErjqmjovjg+WZLFiyn0dvi0eraXzDx4Y9Z8jYks3gPhGMT41p1LU6hvtxX3h37hzWhR2H\n9GzYc4ZPVmTyqUpFQpdgrk+IpHenoAYv3FhjsfH3b/ZyONfI1HE9GejE5MdrjSQWIa5BFqudxeuP\n8ePOU3QI8+XhtN5OD98d2CucaouNRT8c5oPlmTx0S69GxbL3eBGLfjhM79gg7h/VvckWTPRw05Aa\nH0FqfAQ1Ciz9+Rhb9uWx+2ghgX4etcOW+0Sgu8TSKBdjtth4+9u9HDpZwpRxcQySpHJRkliEuMbk\nFVXyj6UHyNFXMCK5PROGdsFNW79P70P7RlFdY+Pr9cfwcNfw9H39GxTLyfxy3luyn/Y6H2ak926S\n2s/FtA/1484bu3Db9Z3Yc6yQjXvzWPFLNhlbsonrGMj1CZEkdQvBTXvppr1zSeVgdgn/MzaO63rX\nr7nuWiKJRYhrhKIobNmfz2drjuCmVTPrjj707RLS4OuNHtCBarOVZZuzCVy2n/TrOtartlFoNPHm\n4j34emn5w4SEi/brNDWtRk2/7qH06x5KcVk1m/blsXFPHv9YdgAfTy2DeodzfZ9I2ofWnbthsdp4\n57t9ZGaX8ODNV24yvNZJYhHiGmCqsfLpmsNsPVBA9+gApt/Si8AmWPE2bXAs1WYbyzeeAJudW6/v\n5NR5ldUW3li8B7PVzlN3JTVJLPUV5O/JLamxjLsuhoPZJWzYc4afd5/mxx2n6BTpz/UJkfTvEYpW\no+Lt7/axP6uYB8f0qPfAgmuRJBYhrnJZeWX8Y+kBDKUm0ofEMm5QjGO59sZSqVRMHNYF1GqWb8nG\n00PDmAEdL3uOxWrn7W/3YTCa+OOdfYnSNXzV8aagVqnoFRtEr9ggyqvM/HKggA17zvDxqkP8+8ej\nhLTz5HRhJZPH9GBIQv0Wvr1WSWIR4ipUXmVm99FCdh42kJldTDtfd2ZPSqJbdECT30ulUjHzjgSM\nZSYWrz+Op7uWGxMvvm67XVH4cEUmR3KNTL+lJz06BjZ5PI3h5+3OTf2jGZncnhNnytiw5wy/HSvk\ngdHduV6SitMksQhxlSitqGHXEQM7Dhs4nGPEriiEtPNkZP9obh7Ysdlmy0PtbohTx/Wkxmzjs9WH\n8XTTMKj3hSOmvvn5ONsP6rljaGcG9nTdEVUqlYrOUe3oHNWutUNpkySxCNGGFZdVs/OwgR2H9Rw7\nVYoChAd5M2ZgB5K7h9IhzLfF9jvXatTMSO/Nm4v38OGKg3i4a+qs8Ltu5yl+2JbDjYlRjBnQoUVi\nEq1DEosQbYy+pOpsMjGQlVcGQHudD7cMjiW5u47IEJ8WSya/5+6mYdYdfXjty994f+l+/nBHAr1i\ng9h9xMAXPx6hb5cQJo3s2mrxiZYhiUWINuBMYSU7D+vZedhAjr4CgJhwP26/oRPJ3UNbfK+Uy/F0\n1/L4nQnM/3w3b3+3lztu6Mw3Px8nJtyPh27p1eAZ76LtkMQixGWUlNfw68ECYiL86RLVrslGU12J\noijk6ivYedjAb8eLyC2o3aujS1Q7Jg7rQr9uOkLqMWO8pfl4uvHkXX159fNdfPHjUXQBnvzhjgQ8\n3JtubTHhuiSxCJd2Sl9Bxi/ZjBnQscX32th+sIBPVx+msrp2JV9fLzcSugST2FVHr9igK24uVV+K\nopCdX86OQ7U1E73RhEoFvTuFcH2fCJK66VplvkdDtfNx5+m7+rJscxZjBnTEv5n3cRGuQxKLcFmV\n1Rb+/u1eCkur2XHIwOgBHbglNaZJV9S9mKpqK5+vPcwvBwqIjfDn/lHdKSip4rejhew6Usjmffm4\nadX0igmib9fazaca+qZpVxSOny5lxyEDu47oKSqrQaNWEdcxkDEDO5DYVUfnmGCX3F3QGUH+nkwe\nE9faYYgWJolFuCS7ovDB8kxKymt4fEICOw/rWbn1JDuPGHhwTI9mmY8BcDinhH9lZFJSbuaW1BjG\nXReDVqOmY7gfKXFhWG12Duca+e1IIbuPGfjtWCEqoHP7diR2DSGxq+6Kizna7HaO5BjZccTAriMG\nSivMaDUqescGkz6kEwldQpp1aLAQzU0Si3BJGVuy2Xu8iHtGdqNP52D6dA4mpWcYn6w6xKuf72JY\nUhS339C5ydaXsljtfL/xBKu35aAL9OLZ+5LoHHnhHAatpram0ismiEkju5JTUMHuowZ+O1rI4vXH\nWbz+OBHB3vQ9m2Q6RfqjVqmw2uwcPFnCzsN6dh0ppMJkwV2rJr5zMP2660joHNIia2UJ0RLkL1m4\nnP0nili6MYuBvcIYlvTfGdy9YoJ4aUoK3204wbodp9hzrJAHRvegd6fgRt3vZF4Z8xftIFdfwdC+\nkUwc1tWpTmaVSkXHcD86hvuRPqQThaUmfjtayO6jhazZnsuqrTn4+7jTKcKfI7lGqmqseLprSOgS\nQr9uOuI7BUtntrgqSWIRLqXQaOIfyw4QpfPhgVE9Lpjv4OmuZdKIbqTEhbFw5UFe/3oPqb3DmTi8\na72bj+yKwo+/5vLthhN4uWuYdXsf+nZt+Gq/Ie28GJEczYjkaCqrLew7XsTuo4Vk5ZWR2DWEft1D\n6RUbeNml2YW4GkhiES7DYrXx7pL92BWFR26Nv+yn+S5R7XjxwRSWb8lm1daT7Msq5t6R3UjuEerU\nvYrLqvlwxUEOniwhpWc4k4Z3adJRSz6ebgzsFS67C4prkiQW4TI+X3uUk/nlPHZbvFMT/ty0am67\nvhPJ3XUsXHmIBUv206+bjntu6kbAZfY235ZZO4zYZld4YHR3bh/RncLCiqZ8KkJc0ySxCJewcc8Z\nNuw5w9hBHUk8b30pZ3QI8+P5B/qxensuSzZmcfCDbdw1vCup8eF1mtKqqi18tuYIWzML6Bzpz9Tx\nPQkL9JblRYRoYpJYRKs7mV/Op2uOENcxkFuHOLdR1O9p1GpuHtiRpG46Pl55kI9WHmTbwQIeGNWd\nkAAvDp4s4cMVmRjLzaQPiWXsoI6ytIgQzUQSi2hVFSYL736/Dz9vNx5K69XoJVPCg7x55p4kft59\nmsU/H+dPH26nT+dgdhzSExrkzXP39yM2wr+JohdCXIwkFtFqzp8EOefeJPy9m6bzXK1SMSypPQmd\nQ/hk9SF+PaTnxsQo7ryxiwzvFaIFSGIRrWb55mz2nSjivpu6XXQyYmMFt/PkiQkJlFWaaXeZznwh\nRNOSRmbRKvadKGLZpiwG9Qpn6CW2sW0KKpVKkooQLUwSi2hxBqOJfy47QJTOl/tHd5dRWUJcZSSx\niBZlsdpY8P1+7Ao8elvvJl96XgjR+lqsjyUrK4s5c+ZgNBoJCAhg/vz5xMTE1Cnz9ttv88UXXxAa\nWjt7Oikpiblz5wIwZ84ctmzZQmBgIACjR49mxowZLRW+aCKfrTnCyYJyZt3eh9BA19n1UAjRdFos\nscydO5dJkyaRlpbG0qVLeeGFF1i0aNEF5dLT05k9e/ZFrzF9+nTuvffe5g5V/I5dUVA3QXPVhj1n\n2Lg3j3HXdWzUmlxCCNfWIomlqKiIzMxMFi5cCMC4ceN4+eWXKS4uJigoqCVCEPVUVW1lx2E9m/bl\nceJ0GbpAL6J1PrQP9SVa50tUqC8h7TydTjjZ+WV8tuYIvWICSR/csEmQQoi2oUUSS15eHmFhYWg0\nte3pGo2G0NBQ8vLyLkgsK1asYNOmTeh0Oh577DESExMdjy1cuJCvvvqK6OhonnzySTp37lyvOIKD\nfRv8HHS6lt0Wt76aIj6bXWHvUQPrfs3ll/15mC02onS+jBsSi6HERPaZMnYcNjjKe3lo6BjuT0xk\nO2LC/YiJbEfHCP8LVhkuqzTz/tIDBPh58OyDA1xylNa18Pttbq4eo8TXclxqHstdd93Fww8/jJub\nG5s3b2bmzJmsXLmSwMBAnnjiCXQ6HWq1miVLljB16lR+/PFHR7JyRlFRBXa7Uu+4dDo/l94atrHx\n5RdXsXlfHlv251NSXoO3h5breoeTGh9Opwj/OqO2qs1WThdWckpfwSl9JbmGCjbsOsUPNVZHmWB/\nD9rrfGkf6kt7nS/bD+kpLqtmzj39MJvMGEzmRj3fpna1/35bgqvHKPE1jFqtatAH8hZJLBERERQU\nFGCz2dBoNNhsNvR6PREREXXK6XT/XXwwNTWViIgIjh49SkpKCmFhYY7H0tPTeeWVV8jPzycqqvnm\nQFzNqqqtbD9UwJZ9+Rw7XYpKBb1jg5k4rAuJXUMuuWeIp7uWzpHt6kxoVBSFkvIaThkqyNVXcMpQ\nySlDBfuzirGdTeT3j+pOp0hZSkWIa0GLJJbg4GDi4uLIyMggLS2NjIwM4uLiLmgGKygocCSQgwcP\ncvr0aWJjYy94bOPGjajV6jrJRlyZ3a6QmV3M5v357DpiwGK1Exniw4QbOzOwZziBfg1rolKpVAT5\nexLk70mfzv/tlLdY7eQVVeLu6U6Yf9PtdSKEcG0t1hT24osvMmfOHBYsWIC/vz/z588HYNq0acya\nNYv4+Hhef/11Dhw4gFqtxs3Njb/+9a+OWszs2bMpKipCpVLh6+vLe++9h1brUi15LiuvqJLN+/L5\n5UBtU5ePp5YhfSJIjY8gJtyv2SYoumnVdAjzc9lqvhCieagURal/p0MbdS32sazbeYrP1x5BrVIR\n3ymI1PgIErqE4KZtubmxbfn1cwWuHh+4fowSX8O4dB+LaB1HTxn5ct1REjoHM3lMD5ccjSWEuPpI\nYrlKlVWaeW/JfoLbeTJtfC9QYYhuAAAgAElEQVS8PeVXLYRoGbJW2FXIblf4x7IDVFZbmZneW5KK\nEA1kN5sp37Edc96Z1g6lTZF3nKvQkk0nOHiyhAdv7kGHsKtn0pUQLUWx2ynf+guF33+LtaQYAI/o\nDvilDMQvZQBuwcGtHKFrk8RyldlzrJCMLScZ0ieCIX0iWzscIdqcqoOZGBZ/RU3OSTxiYgm9734s\nBQWU/7qNwm+/pvDbr/Hs0hW/lAH49euPtl3Tb1LX1kliuYoUGk38KyOTDqG+3DOyW2uHI0SbUnPm\nDIXffEXl3j1og4IJn/YQfv0HoFLX9hgEjhyF2aCnfPs2yrdvw/DFZxj+/TnecT3xSxmIb1ISGm+f\nVn4WrkESy1XCYrWzYEntPiczb4vHXfY5EcIp1tJSipYtoXTjf1B7eBBy+50EjBiB2u3CSb3uulCC\nx44neOx4ak6fOptktlLw8YfoP/sE797x+KcMxCehL2qPa3cUpiSWq8S/1x0lO7+cx26PJzTAq7XD\nEcLl2WtqKFm7muJVK1GsFgKGDiN4fBoaP+f6JT2i2uNxa3uC02+jOiuL8u1bKf91O5W/7Ubl4YFv\n30T8+g/Ap3c8qhaazK0oCpZCAzVZWVRn1/4zFxQQOfNRvDp3aZEYQBLLVWH9zlx+3n2aMQM6kNhV\nd+UThLiGKXY7Zb9soWjJt1hLSvBN7EfI7RNwDw9v0PVUKhVenTrh1akTujvvwnTkcG1NZuevlG/b\nitrbG4/20biFhuKmC8U9NMzxvca7cZvdWUtLaxNI1glHIrFXVNTGpdXi0aEDfv374x4ecYUrNS1J\nLG3cKUMF7yzeQ/foAG67QfY5Ea5Dsdup2LWD0k0bUcyNWNFapUIbFFT7hqzT4RYahrsuFLWvb72X\nI6o6mInh6y+pyc3BIyaW8GkP492te8Nj+32oajXePeLw7hFH6KR7qczcT8WuXZjz86jctxdbaWmd\n8hpfP9xCdZREt8fmH4j72YTjFhqGxq/ucku2qipqTmY7Ekh11gmsxcVnb6zCPTIK376JeMZ2wjMm\nFo+o9i1WU/o9SSxtmKnGyrvf78fHU8vDab3QqGVakmh9is1G+fZtFK9Yjjk/DzedDm1Qw4fnKjYb\npsOHKN/6C5y3ApXay8vxJuweGuqoBbiFhqFt187R6Q5QlZPD6X8upHLfXrTBwYRPfxi/5JQ6ZZqa\nSqvFt09ffPv0dRyzV1djMRgw6wuwGPRY9HosBj1lBw9SYyis8/xUHp64h4aiDQrCUlCAuSDf8bib\nTodXl661CSQmFs+OMS7VpyOJpY1SFIWFKw9iKDHxfzOuk+VaRKtTrFbKtmymeFUGFoMB9/bRRDw8\nE9+k5CZ5A7dbzFgMhVjOvimbz74p1+ScpGL3TrDZHGVVbm5nk0woKq2Wil07azvmJ0wkYNjwi3bM\ntwS1pyce0dF4REfXOa7T+VFwphhrUWHt8zr73Cz6AiyFhbiFheE3YKCjNqLxbfimhS1BEksbtXbH\nKXYcNjBhaGd6dw5xyQXsxLXBbjaTt2IlOd98j7W4GI+YWCInTsKnT0KT1gjUbu54REbiEXnh/CzF\nZsNSXFT7hnz2Tbm2VmDAWmok4ubReI+42aXfkNVubriHR7R4f0hzkMTSBh09ZWTx+mMkdg1h9IAO\nrR2OuEbZq6sx/mc9JWt+wFZaimeXroTd/yDevXo321YMl6LSaHDXheKuC4VeFz7uqqsHX60ksbQx\n5/aPD/b3ZMrYuBb/DyyEraoK408/UvLjGuwVFXjH9aTTM09SExotf48CkMTSppxbXLLCZOG5+/rh\n7enW2iGJa4itooKSH9dgXLcWu8mET58EgsaOx6tzF9pJjUCcRxJLG7JkU1bt4pJjZHFJ0XKspaWU\nrPkB488/odTU4JvUj6Cx4/HsGNPaoQkXJYmljdh7vJCMLdkMjo9gSIIsLimurDrrBMb16yjfuRPF\namn4hex2APxSBhB08zg8oto3UYTiaiWJpQ0oNJr4YHkm0aG+3HuTLC4pLs1uNlP+6zaM63+iJjsL\nlYcnfv37o/Vv+Aq8Kq0Wv5SBDZ6ZLq49klhc3H8Xl1SYeWtvWVxSXJTZoKf0558o3bQRe2Ul7pGR\nhE66F79BqWi8ZO040bIksbiwM4WVfLgik+z8ch69LZ6wwMatKySuLordTuX+vRh/+omqA/tApcI3\nqR8BQ4fh1b2HjNASrUYSiwuyKwo//prLtxtO4OGmYWZ6b5K6yeKSopatvJzSTRsp/c96LIUGNO0C\nCBp3CwE3DEUbENja4QkhicXVFBpNfLjiIIdzjSR0DmbymB6yXEsD2atNVGdnYy0uxrtnzzb9pqso\nCtVZWZSuX0f5r9tQrFa8uvcg5I4J+PZNarXFBoW4GPlrdBGKorBxbx7/XncUFfDgmB4M7hMhzRlO\nslssmE/l1lk+3JyX999F/VQqvLp1d2wn21JLe1jLy6jOOgE2e4OvYbPXcGrlampOZqPy8MR/yA0E\nDB2GR1RUE0YqRNNxKrF88sknjB8/nqCgoOaO55pkrKjh41WH2Hu8iB4dAvifsXGEtJMO10tR7HbM\neXlUZ59NIllZ1OTmOBYh1Pj54RnbCb/kFDxiYtEGBFC55zfKt21F/+kn6L/4DJ+evfBLGYBvYhLQ\ndHOCFKsV04njVO3fR+WB/dTknKyzYm1DuUdGEXrPffgPug61p/xtCNfmVGLZunUrb775JikpKaSl\npTFixAjc3VtnddCrzbbMAj5bcxiz1c7dI7oyvF971FJLcbBbzFiNRmqys2sTSVYW1SdPotRUA2dX\ni+0YQ+DIUXjGxOIZG4s2KPiCmp5nh44EjbuFmtwcx57llR9+gMrNjZL+ybgn9MOnT58GrXprNuip\n2r+fygP7MB06iL26GtRqPDt1JviWdLy690Dt6dng1yA4NIAKD3+pvYo2Q6Uozn2cKikpYeXKlSxb\ntowTJ05w0003kZ6eTv/+/Zs7xiZTVFSB3V7/T4/NsYBdhcnCp6sP8+shPZ0i/ZkyNo6IYJ8GXcvV\nF9gLCfGlINeAvbICW2UltspK7JWV2KrOfj173F5Z9d/vq2rLnb9BlEqrxSO6Q+3+EzGxeMZ2wj08\nvEEr6Cp2O9XHj1O2fStVu3ZgKS1F7emJb2I//AYMwLtHz0v2W9irq6k6dJDKA/upOrAfi74AAG1w\nMD694vHu1RvvuDg03g37ff6eq/9+wfVjlPgaRq1WERxc/2ZjpxPL+Q4dOsQzzzzD0aNHiYiIYMKE\nCdx///34+DTNf6Tm4iqJZc+xQj5edYgKk4W0wbGMGdihUZt0ueofpbXUSNGyJZRt3oRitV6ynMrd\nHY2PD2pvn9qvPrVfHcf8/PDs0BGP9tHN0kkdEuTNyY3bKd++jYpdO7CbTGh8/fDtl4zfgIF4delK\nzalcR/OW6dhRsNlQubvj3b0H3r3i8endG7ew8GapVbjq7/d8rh6jxNcwLZJYfvnlF5YtW8a6devo\n3bs36enpREZGsmjRIgoLC/niiy/qHUBLau3EYqqx8u91R9m0N4/2Oh+mjuvZJGt+udofpb2mhpK1\nqyletRLFaiH0xqHYA0PQ+PheNHGoW7lZ9fzXz26xULV/b22S2fMbitmMSqt1JEb39tH49OqNT+94\nPLt0Re3W/AuButrv92JcPUaJr2Eamlic+vg3f/58VqxYgZ+fH2lpaSxfvpywsDDH4wkJCaSkpNT7\n5teSQydL+HDFQYrLq7l5YEfSBsfipr26thJW7HbKftlM4fffYjMa8U3qR8jtE4jq3dUl/9NcjNrN\nDd/Efvgm9sNeXU3Fnt2Yjh3FM6YTPr16ow0IaO0QhXB5TiWWmpoa3nnnHfr06XPRx93c3Pjmm2+a\nNLCrhcVqZ/HPx/hxxylCA7149p5+dGnf8HWb6sNusVCx81eqT5zAu3dvfHr2brb5DpWZByhc/CU1\nubl4xnYi8qGZeHVt2+uaqT098R8wCP8Bg1o7FCHaFKfeZR566CE8fzeqpbS0lOrqakfNpXPnzpe9\nRlZWFnPmzMFoNBIQEMD8+fOJiYmpU+btt9/miy++IDQ0FICkpCTmzp0LgMlk4tlnn+XAgQNoNBpm\nz57NjTfe6NSTbE0b957hxx2nGJYUxYShXfBwb/61vixFRZT+Zz2lG/+DrbwcNBqMP/2I2scHv379\n8UsZgFe37k2ybWzN6VMYFn9N1f69aENCiJg+A9/+KTKCSYhrmFOJZebMmfzlL3+hXbv/ftLOz8/n\n+eefZ/HixU7daO7cuUyaNIm0tDSWLl3KCy+8wKJFiy4ol56ezuzZsy84/uGHH+Lr68vatWvJzs7m\nnnvuYc2aNS4/YOC0oRIfTy333tS9We+j2O1UHczEuH4dlXt+A8AnoS8BNw7Hq1s3qg4coHz7Nsq2\nbqF0w89oAgLwS07BL2UgnrGx9U4E1lIjRUu/p3TjBtReXoRMmEjAsBEt0ucghHBtTiWWrKwsunev\n+8bYvXt3Tpw44dRNioqKyMzMZOHChQCMGzeOl19+meLiYqcnXa5atYpXX30VgJiYGHr37s2GDRsY\nM2aMU+e3Fr3RhC6g+Sa02SorOb3lZ06vWIWloACNnx9BY8bS7oahuAWHOMr59k3Et28i9poaKvf8\nRtn2rZT+/BPGH9fgpgutnZGeMuCKe23Ya2ooWfMDxT+sRLFaCRg+kuBxt7TYTHYhhOtzKrEEBwdz\n8uRJOnbs6Dh28uRJApzsyMzLyyMsLAyNprYZSKPREBoaSl5e3gWJZcWKFWzatAmdTsdjjz1GYmIi\nAGfOnCHqvCUsIiIiyM/Pd+r+rclQYiImoul3e6w+mY1x/U+Ub9+KYjbj2bkLwbek45uUfNlag9rD\nw5FEbFWVVOzaRfn2rRSvzKB4xXLco9o7HnfXhTrOU+x2yrZspnDJ2Y75fsmE3DYB9/MGcQghBDiZ\nWG6//XYee+wxnnjiCaKjo8nJyeGtt95iwoQJTRrMXXfdxcMPP4ybmxubN29m5syZrFy5ksDAplk8\nsCHD5s7R6eqfHKw2O0Vl1dzQr32Dzv89u9lM4eYt5K9aTfnhI6g9PAgdej3hY0bj2ym2AVf0g443\nw603YzYaKdq8BcOGTRR9/y1F33+Lb9eu6K4fjEeojpx/f0VV9kn8uncjZs7T+Mf1qNedmuL5NyeJ\nr/FcPUaJr+U4lVimT5+OVqtl/vz55OfnEx4ezoQJE3jwwQeduklERAQFBQXYbDY0Gg02mw29Xk9E\nRESdcjrdf5eGT01NJSIigqNHj5KSkkJkZCSnT5921HDy8vIYMGCAs88TaPl5LPqSKmx2BV93TaOG\n21oKDRh/Xk/Zpo3YKspxCwtHd9ck/K9LRePtg2+TjIHXoE0ZQkTKEEKKCinfvp3y7VvJ+rC2+dIt\nREfEQzPxTe5PjUpVr/u56hj9cyS+xnP1GCW+hmnWeSxqtZqpU6cyderUet8AapvS4uLiyMjIIC0t\njYyMDOLi4i5oBisoKHCMMjt48CCnT58mNrb2k/jo0aP56quviI+PJzs7m3379vHaa681KJ6Wojea\nAAgNvHIfi72mBotej9mgx6IvwGLQO362FhUB4NM3kYAbh+Md17NZR125BYcQNOZmgsbcjDnvDDVn\nzuDTJ0E65oUQTnF6UoPZbCYrK4uSkhLOn6w/aJBzY/xffPFF5syZw4IFC/D392f+/PkATJs2jVmz\nZhEfH8/rr7/OgQMHUKvVuLm58de//tVRi5kyZQpz5sxh5MiRqNVqXnrpJXxdvMPYUFKbWM513tsq\nK7HoC84mj7P/DHrMej22UmOdc9W+vriHhuLVuQvuQ27Af9B1uAUFt/hzcI+IxD0issXvK4Rou5xa\n0mXHjh08/vjjmM1mKioq8PX1pbKykvDwcNatW9cScTaJpmwKU2w27FVVtYsq/n4xxaraxRRPHMun\nxFBCzxAtFoMBe1VlnWtoAgJwDw3DTReKW2joed/r6rWAoatWo8+R+BrH1eMD149R4muYZm0Ke+WV\nV5g6dSqTJ0+mf//+bN++nXfeeQcvr6t/XwizQc/hRR9SZSjGVlnhSBp2k+my56m9vPDEjQCVOxrf\ncDxjO+Gm09Umj9BQ3EJ0qD1kZ0ghxNXHqcSSnZ3N/fffX+fY9OnTGT58OFOmTGmWwFyFrbwCU24u\ndrUWbUAA6sjICxZTPLcqr+O4tzcqjYYXPtxGsL8ngyYktPbTEEKIFuNUYvHz86OiogJ/f390Oh3H\njh0jICCAqqqq5o6v1Xl16kTfN1+rdzVVURQMxmp6dGy7+6wLIURDOJVYRo4cyX/+8x/Gjx/P7bff\nzv33349Wq2XUqFHNHV+bVVZppsZiIyzQu7VDEUKIFuVUYnnuuecc30+ZMoWEhAQqKysZMmRIswXW\n1p0batycy7kIIYQruuLytjabjREjRmA+b4vY5ORkbrjhBtRNsDru1Upf4vwcFiGEuJpcMTNoNBo0\nGg01NTUtEc9Vw2A0oVJBSDvPKxcWQoiriFNNYffffz+PP/44Dz30EOHhdff1jo6Obrbg2jK90USQ\nnydajdTqhBDXFqcSy8svvwzA5s2b6xxXqVQcPHiw6aO6ChhKTNIMJoS4JjmVWA4dOtTccVx19EYT\niV11Vy4ohBBXGWmnaQamGivlVRapsQghrklO1VgmTZp0ydV0P//88yYN6GrgGBEmQ42FENcgpxLL\n7zf0MhgMfPvtt4wfP75ZgmrrDDKHRQhxDXMqsdx6660XHBs1ahTPPvssjz76aJMH1dbVZx8WIYS4\n2jS4jyUsLIzDhw83ZSxXDX2JCV8vN7w8nN7uRgghrhpOvfN98803dX6urq5mzZo19O3bt1mCausM\nRhlqLIS4djmVWJYuXVrnZ29vbxITE5k8eXJzxNTm6UtMdI1u19phCCFEq3AqsXz66afNHcdVw2qz\nU1xeTWhAeGuHIoQQrcKpPpYlS5ZcMEny0KFDLFmypFmCassKS6tRFBkRJoS4djmVWN566y0iIiLq\nHAsPD+ett95qlqDaMlnVWAhxrXMqsVRUVODr61vnmJ+fH2VlZc0SVFt2bg6LTI4UQlyrnEosnTt3\nZvXq1XWOrV27ls6dOzdLUG2ZvsSEh5sGfx/31g5FCCFahVOd90899RTTp09n1apVREdHk5OTwy+/\n/MI///nP5o6vzTEYTegCPC+5BI4QQlztnKqxJCcnk5GRQXx8PCaTiT59+pCRkUG/fv2aO742R280\nSce9EOKa5lSNxWw2o9PpmD59uuOYxWLBbDbj7i5NPufYFQV9iYn4TkGtHYoQQrQap2osDz74IAcO\nHKhz7MCBA0yZMqVZgmqrjOU1WG126bgXQlzTnEosR44cISEhoc6xPn36yAZgv+NY1ViGGgshrmFO\nJRY/Pz8KCwvrHCssLMTLS95Azyf7sAghhJOJ5aabbuLJJ5/kyJEjmEwmDh8+zDPPPMPo0aObO742\nRW80oVapCPL3bO1QhBCi1TiVWJ544gk6d+7MhAkTSExMZOLEiXTu3JnHH3+8ueNrUwxGE8HtPNBq\nZMdnIcS1y6l3QA8PD+bOnctvv/3Gli1b+PLLL3F3d+emm25q7vjaFH2JidBA79YOQwghWpXTO1EV\nFxezfPlyx4KUycnJPPfcc80ZW5tjMJqIjfBv7TCEEKJVXTaxWCwWfvrpJ77//ns2bdpEhw4dGDt2\nLKdPn+bNN98kODjY6RtlZWUxZ84cjEYjAQEBzJ8/n5iYmIuWPXHiBLfeeiuTJk1i9uzZAMyZM4ct\nW7YQGBgIwOjRo5kxY4bT929uldUWKqutMjlSCHHNu2xiSU1NRaVScdttt/HYY4/Rq1cvAP7973/X\n+0Zz585l0qRJpKWlsXTpUl544QUWLVp0QTmbzcbcuXMZMWLEBY9Nnz6de++9t973bgmyqrEQQtS6\nbB9L9+7dKS8vZ8+ePezbt4/S0tIG3aSoqIjMzEzGjRsHwLhx48jMzKS4uPiCsv/85z8ZOnToJWsz\nrkpWNRZCiFqXTSyffvopa9euJTU1lY8++ojU1FQefvhhqqqqsFqtTt8kLy+PsLAwNBoNABqNhtDQ\nUPLy8uqUO3ToEJs2bbrklscLFy5k/PjxzJw5k+PHjzt9/5ZwrsYiTWFCiGvdFTvvo6KieOSRR3jk\nkUfYsWMHS5cuRa1Wc8stt3D77bfzzDPPNEkgFouFP/3pT7zyyiuOBHS+J554Ap1Oh1qtZsmSJUyd\nOpUff/zxomUvJTjY98qFLkGn87vs4+XVVgL9PGgfFdDgezTGleJrbRJf47h6fOD6MUp8LcfpUWFQ\nu8pxcnIyzz//PGvXrnV6a+KIiAgKCgqw2WxoNBpsNht6vb7OrpQGg4GcnBzHQpdlZWUoikJFRQUv\nv/wyYWFhjrLp6em88sor5OfnExUV5XT8RUUV2O2K0+XP0en8MBjKL1smJ6+M4HaeVyzXHJyJrzVJ\nfI3j6vGB68co8TWMWq1q0AfyeiWWczw8PBg3bpyjz+RKgoODiYuLIyMjg7S0NDIyMoiLiyMo6L+r\nAEdGRrJt2zbHz2+//TZVVVWOUWEFBQWO5LJx40bUanWdZNPa9EYTcR0DWzsMIYRodQ1KLA3x4osv\nMmfOHBYsWIC/vz/z588HYNq0acyaNYv4+PjLnj979myKiopQqVT4+vry3nvvodW2WPiXZbbYKCmv\nkY57IYSgBRNL586dWbx48QXHP/jgg4uWf+yxx+r8/PHHHzdHWE3CUFoNyKrGQggBTi7pIi7PIKsa\nCyGEgySWJqCXfViEEMJBEksTMJSY8PLQ4Ofl1tqhCCFEq5PE0gT0RhO6AC9UKlVrhyKEEK1OEksT\n0BtN0r8ihBBnSWJpJLtdodBokv4VIYQ4SxJLIxWXV2OzK1JjEUKIsySxNJIMNRZCiLoksTSSDDUW\nQoi6JLE0kt5oQqNWEeTn2dqhCCGES5DE0kiGEhMhAV6o1TLUWAghQBJLo8lQYyGEqEsSSyMoioK+\nRBKLEEKcTxJLI5SbLFSbbdJxL4QQ55HE0ggy1FgIIS4kiaURZKixEEJcSBJLIxhKTKiA0AAZaiyE\nEOdIYmkEvdFEgJ8HblpNa4cihBAuQxJLI8hQYyGEuJAklkYwlMiqxkII8XuSWBqoxmyjtNIsNRYh\nhPgdSSwNZDg7IixUaixCCFGHJJYGcgw1lhqLEELUIYmlgfQlUmMRQoiLkcTSQAajCR9PLT6ebq0d\nihBCuBRJLA2kN5qkGUwIIS5CEksD6UuqpBlMCCEuQhJLA1htdopKa6TGIoQQFyGJpQGKy6qxK4rM\nYRFCiIuQxNIAepnDIoQQlySJpQEc+7AEerdyJEII4XoksTSA3mjCTaumna97a4cihBAup8USS1ZW\nFhMnTmTUqFFMnDiR7OzsS5Y9ceIECQkJzJ8/33HMZDLx+OOPM3LkSEaPHs369etbIOqL05fUDjVW\nq1StFoMQQriqFkssc+fOZdKkSaxevZpJkybxwgsvXLSczWZj7ty5jBgxos7xDz/8EF9fX9auXcv7\n77/P888/T2VlZUuEfgGDLJcvhBCX1CKJpaioiMzMTMaNGwfAuHHjyMzMpLi4+IKy//znPxk6dCgx\nMTF1jq9atYqJEycCEBMTQ+/evdmwYUOzx/57iqJgMFbLUGMhhLgEbUvcJC8vj7CwMDSa2p0WNRoN\noaGh5OXlERQU5Ch36NAhNm3axKJFi1iwYEGda5w5c4aoqCjHzxEREeTn59crjuBg3wY/B53OD4CS\nsmpqLDY6RQc4jrkCV4rlYiS+xnH1+MD1Y5T4Wk6LJBZnWCwW/vSnP/HKK684ElBTKyqqwG5X6n2e\nTueHwVAOwNFTRgC8tGrHsdZ2fnyuSOJrHFePD1w/RomvYdRqVYM+kLdIYomIiKCgoACbzYZGo8Fm\ns6HX64mIiHCUMRgM5OTkMH36dADKyspQFIWKigpefvllIiMjOX36tKOGk5eXx4ABA1oi/DpkVWMh\nhLi8FkkswcHBxMXFkZGRQVpaGhkZGcTFxdVpBouMjGTbtm2On99++22qqqqYPXs2AKNHj+arr74i\nPj6e7Oxs9u3bx2uvvdYS4ddhMJpQqSCknWeL31sIV2GzWSkpMWC1mgHQ69XY7fZWjurSJL7L02rd\nCQzUodE0TUposaawF198kTlz5rBgwQL8/f0dQ4mnTZvGrFmziI+Pv+z5U6ZMYc6cOYwcORK1Ws1L\nL72Er2/D+0waSm80EeTniVYjU4DEtaukxICnpzc+PuGoVCq0WjVWq+u+cUt8l6YoCpWVZZSUGAgJ\nibjyCU5QKYpS/06HNqop+ljmLdqBh5uGp+9ObOrwGsxV22fPkfgaxxXjy88/SVhYB1Rn53LJG3fj\ntHZ8iqJQUJBDeHjHOscb2sciH7vr6dzkSCGudSqZIHzVaOrfpcuMCmsLTDVWKkwW6bgXwoVMm/YA\nFosFq9VCbm4OsbGdAejWrTv/+79z63WtP/7xUZ5++n+JiIi8bLm//OXPjB+fTnx8QoPjvppJYqkH\nx4gwqbEI4TI++OATAPLyzjB16n18/PEXlyx7bmTqpbz++jtO3bO+CetaI4mlHgyyXL4Qbcqvv25j\nwYK36NKlK0eOHOHhhx+ltNTIt99+hdVqRaVS8eijT5CUlAzArbfezJtvLqBjxxhmzJhCfHwf9u3b\nS2GhgZEjRzN9+kwAZsyYwgMPTGHgwOt46aU/4e3tw8mTWej1BSQkJPLssy+gUqkoKMhn3ry5lJSU\n0L59e2w2G6mpQ0hPv6NOnGazmSef/AOlpaWYzWZ69erN00//L1qtFkVRWLToI9atW4NKpcbLy4v3\n3/8IgOXLl/DNN18B4Obmxv/7f28RGBjYgq/wxUliqYdz+7BIH4sQ/7V5Xx6b9+XRHMOABveJIDW+\ncSOVjh8/xpw5z9O9e08ASkuNjB49FoCsrBM8+eRjfPfdioueq9freffdD6isrOTOO9MYNy6NyMio\nC8plZ59w1HYmT76b3cV4DxYAABO5SURBVLt3kpSUzBtv/JWUlEHcd99kzpw5zQMP3E1q6pALztdq\ntfz5z3/B378ddrudl176E6tWZTB+fDorVixl69YtvP/+R3h7+2A01k7S/vXXbXz++SLee+9fBAYG\nUVVViVbr1qjXqqlIYqkHfYkJP283vDzkZROirejYMYZevXo7Rl3l5uby4ovPUVhoQKPRUlhowGg0\nEhAQcMG5w4bVTm/w8/OjQ4eOnD596qKJ5frrh+LuXruNRteu3Tl9+hRJScns2rWTZ555DoDIyCgS\nE/tdNEa73c5nn33C9u1bsdttlJWV4e/vD8DmzZu49dYJeHv7ADji/OWXTdx88zgCA2vnA5573BXI\nO2Q9yKrGQlwoNT6CGxKjXHY4r5dX3Q355s59lj/+cTapqUOw2WwMH56K2Vxz0XPPJQsAtVqNzWa7\nYrna1UWs9Yrxhx9WcPDgARYs+Bfe3t4sXPgBBQX1WwvRlchw43rQl5jQSf+KEG1aZWWFY9TX8uXf\nY7XWLwnUR2JiEqtWZQCQn5/H7t07L1quvLycdu0C8Pb2pqysjB9/XO14LDV1MN9/v5iqqioAR1PY\nddcNYeXKDEpKaleJr6qqxGw2N9tzqQ+psTjJarNTXF5NaEB4a4cihGiEWbOeZPbsJ/Dz82PQoMHN\nuoLHE0/MZt68F1i1agWRkVH07NkLH58L7zd27C1s3LiBSZNuJygomISERMcSL2PHplFYWMj06ZPR\narV4e3uzYMG/SE5O4e677+UPf5iBSqXG3d2d//f/3qpTe2otMvPeCTqdH/sOF/C//9zKlLFxje5M\nbGquODP7fBJf47hifPn5J+vM0v7/7d17VJT1usDx7zCApsRVQFC85oXSjhdExYAl3jghaBRlHjFF\nLQVvtb0uWprINulszUsUoalHSyvvIuAlOu7YmKlH1868LZcpKCAoF3FARYf3/GG+iYAMOsygPZ+1\nXMvh/fGbZ2aeenzfd36/x9wrx2tjrvhu376FpaUVWq2Wq1fzGT9+NPHxq2jZ0qNBxPeghz9TaOC7\nGz8LZFdjIURdZWZeZNGiGBRFQa/XM2HCpCpF5VkkhcVA6hoWuXkvhDBQx46dH7lg81klN+8NlF90\nk0ZWWmybmv/6pRBCNGRSWAx0tfgmzvaNZeM9IYSohRQWA+UVlcmKeyGEMIAUFgNUVChcLb6Fq0OT\n2gcLIcRfnBQWAxSW3OKuvkIWRwohhAGksBggt6AUkG+ECdEQ/e1vU9mxY0ulnymKQljYsBpXut83\nefK7ZGSkA7B6dQJpafuqHffVV1/y2WfLao0lJSWJrKxM9fG//vVP4uOX1/p7zxr5urEBrly7V1jk\njEWIhicoKIRvv/260lb0x4//HxYWGrp162HwPOPHT3ziWFJSkrCzs6dVq3sLDV95xZ9XXvF/4nmf\nNlJYDJBbUIrWQoOTbSNzhyKEeIivrz9LlnzMxYsXaNOmLQDJybt49dVgNBoNR48eZvXqL7h9+zZ6\nvZ7RoyMYOHBIlXn+/veP6NzZk9dffwudTsfixTH8/vt5HB2dcHV1xcHBCYCjRw+zatUXlJdXni85\neRdnz55m2bJ/sGrVF0RFTePq1XwOHkwnNvYTAL7+eh1796YA4On5EtOnz6RJkyasWpXAxYsXKS3V\nkZOTTYsWLVm4MI7GjRtXiXPBgg/Jysrkzp1yWrTwYO7ceepOyLt372Tz5m+Be/1ZPvnkUxwdncjI\nSGfNmkTu3r2LhYWG6OgFvPBCB+N/GH+QwmKAKwVlONk2RmshVw6FeFjJwQxKMtKpj92h7F7xw9an\n3yPHWFlZMWjQf5KSsovIyGmUlZWSnv5Pvv76e+DeIsUvv1yDomgoLCxg3LhwvL37qv8zrs7atato\n0qQpGzdupbi4mIiI/yIgYJA63+efr0ar1VaaLygohNTU3bz9drjacyUlJUmd8+efM9i7N0XtqxIb\nO59161YTGTkVgLNnT7Nq1XpsbGz44IPJ7NuXSkjIa1VimzZthrp1fmLi53zzzf8wadIUjh07yoYN\na/n889U4OTWjrKwMrVZLVlYmcXGxxMevwsOjFeXl5dy9e6cOn0LdSWExQG5BqVwGE6IBCwoKYcaM\nKbz33mTS0vbTtet/4OLiCkBxcRFxcQvJyspEq7WkpOQ6WVmZdOnStcb5jh8/yvTpM4F7/U/8/QPU\nY8XFRXz8cQyXL2cZPB/cO9MZMGCwugllSEgoy5f/Qz3u7d2H559/HoAXX+xCdvblaufZs2c3+/bt\n4e7dO9y8eQsPj1bAvcIVGBiEk1MzAJo0ufct1iNHfqFPHx91nLW1db1vVCmFxQBXrpXSq7OLucMQ\nokGy9emHo5+vWTdR7NChI05Ozhw6dJCUlF2EhY1Ujy1Zshg/P39iYz9Bo9EwYkRojf1XDLFkyWL6\n9fNj0aL/Nsp891lb/3mpvabeL//+93F27NjKF1+swcHBgX379rBr17Ynfm5jk2s7tSi9dQfdzTuy\nOFKIBi4oKIQ1axK5dCkLX98/b5jfuHEDNzd3NBoNR44cIjv7Uq1z9ejRS72Mdf16MT/99L8PzedW\n7XxNmzaltFRX7ZxeXt78+ON+yspKURSF3bt30KtX7zq9xhs3btC0qQ12dnaUl5eTnLxLPda3bz/2\n7EmmsLAAgLKyMm7fvo23dx8OHTrIpUtZAJSXl1NWVlqn560rOWOphexqLMTTYdCgQOLjlxMS8hpW\nVn/2fp80aTJLlsSRmJiAp+eLtG9f+03rMWPG8/HHC9T+KN26da8y31dfJVaZLyQklM8++5SNGzcQ\nFTWt0px9+/bj/PlzvPfeWAA6d36Rd94ZV6fX2KePD/v2pfL226HY2dnTrVt3Tp06CUCPHl6Eh49h\n+vTIP/qzWBEX9ykeHq2YNSua+fPnotdXoNVaEB29gPbtX6jTc9eF9GOpxeHTeSTsPElMhDctXeqv\nIdCTaIj9Oh4k8T2Zhhif9GMxroYQnzH7scilsFpUKArPN7GSm/dCCGEguRRWi96ergzs0xZdyU1z\nhyKEEE8FOWOphUaj4blGUn+FEMJQUliEEI/lL3R79pln7M9SCosQos4sLa0pLS2R4vIMUBSF0tIS\nLC2Nt2jSZNd4Lly4wJw5cyguLsbe3p64uDjatGlTaczWrVtZt24dFhYWVFRUEBYWxujRowFYuXIl\nGzduxMXl3kLFHj16MH/+fFOFL4R4gIODM0VFV9HpigHU/2YbKonv0SwtrXFwcDbefEabqRbz589n\n5MiRDBs2jJ07dzJv3jzWr19facyQIUMIDQ1Fo9Gg0+kIDg7G29ubzp07AzB8+HBmz55tqpCFEDXQ\nai1p1sxNfdwQvxL9IInPtExyKaygoIBTp04xdOhQAIYOHcqpU6coLCysNM7GxkbtKX/r1i3u3Lkj\nPeaFEOIpY5LCkpubi6urK1qtFgCtVouLiwu5ublVxqalpREUFET//v0ZP348nTp1Uo8lJycTHBxM\nREQEx48fN0XoQggh6qjBfY92wIABDBgwgJycHKKiovDz86Ndu3aMGDGCiRMnYmVlRUZGBpGRkaSk\npODg4GDw3I+zgvQ+Z+fnH/t3TUHiezIS35Nr6DFKfKZjkjMWNzc38vLy1N069Xo9+fn5uLm51fg7\n7u7udO3alQMHDgDg7Oys7v/Tr18/3NzcOHfuXL3HLoQQom5MUlicnJzw9PRk9+7dAOzevRtPT08c\nHR0rjTt//rz698LCQn755Rc6duwIQF5ennrs9OnTZGdn07ZtWxNEL4QQoi5Mtgnl+fPnmTNnDiUl\nJdja2hIXF0e7du2YMGECU6dOpWvXrixatIiMjAwsLS1RFIWwsDDCw8MBmD17NidPnsTCwgIrKyum\nTp2Kv/9fr5e0EEI0dH+p3Y2FEELUP1l5L4QQwqiksAghhDAqKSxCCCGMSgqLEEIIo5LCIoQQwqik\nsAghhDCqBreli7kYsq2/Xq8nNjaW9PR0NBoN7777LmFhYSaJr6ioiFmzZpGVlYW1tTWtW7cmJiam\nyiLTOXPmcPDgQXWrm8DAQCZNmmSSGAMCArC2tqZRo0YAzJgxA19f30pjbt68ydy5czl58iRarZbZ\ns2fTv3//eo/t8uXLREVFqY9v3LiBTqfj8OHDlcaZsj1DXFwce/fuJTs7m6SkJHUxsCG5CPWfj9XF\nZ2geQv3nYk3vnyF5CPWfi9XFZ2gewlPeKkQRiqIoSnh4uLJjxw5FURRlx44dSnh4eJUx27dvVyIi\nIhS9Xq8UFBQovr6+yqVLl0wSX1FRkXLo0CH18eLFi5W5c+dWGTd79mxlw4YNJonpYf3791fOnj37\nyDErV65UoqOjFUVRlAsXLig+Pj6KTqczRXiVxMbGKgsWLKjy8xUrViiLFy82SQxHjhxRcnJyqrxv\nhuSiotR/PlYXn6F5qCj1n4s1vX+G5KGi1H8u1hTfg2rKQ0UxbS4am1wKw/Bt/VNSUggLC8PCwgJH\nR0cGDhzInj17TBKjvb09vXv3Vh9369aNnJwckzy3MaWmpvLWW28B0KZNG7p06cJPP/1k0hjKy8tJ\nSkri9ddfN+nzPszLy6vKfnmG5iLUfz5WF19DysPq4quL+s7F2uJrKHlYH6SwYPi2/rm5ubi7u6uP\n3dzcuHLlikljBaioqGDTpk0EBARUe3zt2rUEBwcTGRlZaf81U5gxYwbBwcF89NFHlJSUVDmek5ND\nixYt1MfmeA9//PFHXF1deemll6o9bs72DHVpMWHufKwtD8F8uVhbHoL5c7G2PISnt1WIFJan0MKF\nC2nSpAmjRo2qcuz9999n//79JCUlMXjwYMaPH6/uKl3fvvnmG3bt2sXWrVtRFIWYmBiTPG9dbd26\ntcZ/JY4YMYK0tDSSkpIYN24ckZGRFBUVmTjCp8Oj8hDMl4vPQh7C052LUlgwfFt/Nze3Sqf9ubm5\nNG/e3KSxxsXFkZmZybJly7CwqPrxubq6qj8fPnw4ZWVlJvtX2P33y9rampEjR3Ls2LEqY9zd3cnO\nzlYfm/o9zMvL48iRIwQHB1d73NztGerSYsKc+VhbHoL5ctGQPATz5mJteQjmz8UnIYUFw7f1DwwM\nZPPmzVRUVFBYWMgPP/zAkCFDTBbn0qVL+e2334iPj8fa2rraMQ+2F0hPT8fCwgJXV9d6j62srIwb\nN+717FYUhZSUFDw9PauMCwwM5LvvvgPg4sWLnDhxotpv7NSX7du34+/vX2ODOHO3ZzA0F8F8+WhI\nHoJ5ctHQPATz5mJteQjmz8UnIbsb/8GQbf31ej0xMTFkZGQAMGHCBPXmX307d+4cQ4cOpU2bNjRu\n3BiAli1bEh8fz7Bhw0hMTMTV1ZUxY8ZQUFCARqPBxsaGWbNm0a1bt3qP79KlS0yZMgW9Xk9FRQXt\n27fnww8/xMXFpVJ8ZWVlzJkzh9OnT2NhYcHMmTMZOHBgvcd335AhQ4iOjsbPz0/92YOfsSnbM8TG\nxrJv3z6uXbuGg4MD9vb2JCcn15iLD8da3/lYXXzLli2rMQ8Bk+ZidfElJCTUmIcPx1ffuVjT5wvV\n5yGYLxeNTQqLEEIIo5JLYUIIIYxKCosQQgijksIihBDCqKSwCCGEMCopLEIIIYxKCosQDVynTp3I\nzMw0dxhCGEy2zReijgICArh27Zq6nxfAa6+9xrx588wYlRANhxQWIR5DQkICPj4+5g5DiAZJLoUJ\nYSTbtm1jxIgRxMTE0LNnTwIDA/n555/V43l5eUycOBFvb28GDRrE999/rx7T6/UkJCQwcOBAunfv\nTmhoaKUdjQ8ePMjgwYPx8vJiwYIF3F/XnJmZyahRo+jZsye9e/dm+vTppnvBQtRAzliEMKJff/2V\nwMBADh06xP79+5k8eTJpaWnY29vzwQcf0KFDB9LT0/n9998ZO3YsHh4e9O3bl7Vr15KcnExiYiJt\n27bl7Nmz6pYpAAcOHGDLli3odDpCQ0Pp378/fn5+LF++nH79+rF+/Xru3LnDiRMnzPjqhbhHzliE\neAxRUVF4eXmpf+6ffTg6OvLOO+9gZWXFq6++Stu2bTlw4AC5ubkcO3aMGTNm0KhRIzw9PQkLC2Pn\nzp0AbN68mWnTptGuXTs0Gg2dO3eutEHhhAkTsLW1xd3dnd69e3PmzBkALC0tycnJIT8/n0aNGuHl\n5WX6N0OIh0hhEeIxxMfHc/ToUfXPm2++CdzbKl6j0ajj3N3dyc/PJz8/Hzs7O2xsbCodu7+D7ZUr\nV2jVqlWNz+fs7Kz+/bnnnqO0tBSAmTNnoigKb7zxBkFBQWzZssWor1OIxyGXwoQwory8PBRFUYtL\nbm4uAQEBuLi4cP36dXQ6nVpc7neLBGjevDlZWVl07NixTs/n7OxMbGwsAEePHmXs2LH06tWL1q1b\nG/FVCVE3csYihBEVFhaq9ztSU1M5f/48/v7+uLm50b17d5YuXcrt27c5c+YMW7ZsISQkBICwsDCW\nL1/OxYsXURSFM2fOGNQtMDU1VW2eZWdnh0ajqbHxlhCmImcsQjyGiRMnVlrH4uPjw4ABA3j55ZfJ\nzMykT58+NGvWjBUrVqj3SpYuXcr8+fPx9fXF1taWKVOmqF9ZHjt2LOXl5URERFBUVES7du3UHieP\ncuLECRYtWoROp8PJyYno6Gg8PDzq50ULYSDpxyKEkWzbto3NmzezadMmc4cihFnJObMQQgijksIi\nhBDCqORSmBBCCKOSMxYhhBBGJYVFCCGEUUlhEUIIYVRSWIQQQhiVFBYhhBBGJYVFCCGEUf0/d6f0\nIi5wzpAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY8AAAEcCAYAAAA/aDgKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot\n",
    "\n",
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    " \n",
    "epochs = range(len(acc))\n",
    " \n",
    "plt.plot(epochs, acc, 'b', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'r', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()\n",
    " \n",
    "plt.figure()\n",
    " \n",
    "plt.plot(epochs, loss, 'b', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'r', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "PlFyKNAFTeTu"
   },
   "source": [
    "# **Maximum Accuracy**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "colab_type": "code",
    "id": "9QEYs1daA3r9",
    "outputId": "41d8a7e1-8254-4c72-aaee-0fe1d60defd9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validation Accuracy: 0.5388813119224116\n",
      "Training Accuracy 0.6034537894793658\n"
     ]
    }
   ],
   "source": [
    "# Max Accuracy\n",
    "\n",
    "print(\"Validation Accuracy:\", max(history.history['val_acc']))\n",
    "print(\"Training Accuracy\", max(history.history['acc']))"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "Project_ResnetInceptionV2-1.ipynb",
   "provenance": [],
   "toc_visible": true,
   "version": "0.3.2"
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}