Models_For_Segmentation.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w5VPRht7-opS"
      },
      "source": [
        "# MODEL 1"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "AGG6wgp00M2Y"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import time\n",
        "from skimage import io\n",
        "import cv2\n",
        "import random\n",
        "\n",
        "import tensorflow as tf\n",
        "'''from keras.models import Model\n",
        "from keras.models import Input\n",
        "from keras.layers import Conv2D\n",
        "from keras.layers import Conv2DTranspose\n",
        "from keras.layers import add\n",
        "from keras.layers import Activation\n",
        "from keras.layers import UpSampling2D\n",
        "from keras.layers import Dropout\n",
        "from keras.layers import BatchNormalization\n",
        "from keras.layers import MaxPooling2D\n",
        "from keras.layers import Lambda\n",
        "from keras.layers import multiply\n",
        "from keras.layers import concatenate'''\n",
        "\n",
        "from tensorflow.keras import backend as K\n",
        "from tensorflow.keras.metrics import MeanIoU\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "98yP7nYm0bIh",
        "outputId": "2f069176-1cf6-4192-a38f-f230ced0b536"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "['/content/drive/MyDrive/Intercubic/TrainingSet/Image', '/content/drive/MyDrive/Intercubic/TestSet/Image']\n"
          ]
        }
      ],
      "source": [
        "import dataset\n",
        "mySet=dataset.Dataset(\"/content/drive/MyDrive/Intercubic\",\"TrainingSet\",\"TestSet\",\"Segmentation\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "jYJcou4IbRTn",
        "outputId": "d445fc7f-b478-401a-c3c2-46a7158d4372"
      },
      "outputs": [],
      "source": [
        "mySet.importSegmentationClass(False)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "QCZEfMIWlo92"
      },
      "outputs": [],
      "source": [
        "mySet.trainingArrayX=mySet.trainingArrayX.astype('float32')\n",
        "mySet.trainingArrayY=mySet.trainingArrayY.astype('float32')\n",
        "mySet.testArrayX=mySet.testArrayX.astype('float32')\n",
        "mySet.testArrayY=mySet.testArrayY.astype('float32')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "K27Mo4-31sxM"
      },
      "outputs": [],
      "source": [
        "\n",
        "for i in range(len(mySet.trainingArrayX)):\n",
        "  mySet.trainingArrayX[i]=mySet.trainingArrayX[i]/255.\n",
        "  mySet.trainingArrayY[i]=mySet.trainingArrayY[i]/255.\n",
        "  '''trainlistx.append(cv2.resize(mySet.trainingArrayX[i],(256,256),0,0,interpolation = cv2.INTER_NEAREST))\n",
        "  trainlisty.append(cv2.resize(mySet.trainingArrayY[i],(256,256),0,0,interpolation= cv2.INTER_NEAREST))\n",
        "  '''"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "wvHBhMgNm2mZ"
      },
      "outputs": [],
      "source": [
        "for i in range(len(mySet.testArrayX)):\n",
        "  mySet.testArrayX[i]=mySet.testArrayX[i]/255.\n",
        "  mySet.testArrayY[i]=mySet.testArrayY[i]/255.\n",
        "  '''testlistx.append(cv2.resize(mySet.testArrayX[i],(256,256),0,0,interpolation = cv2.INTER_NEAREST))\n",
        "  testlisty.append(cv2.resize(mySet.testArrayY[i],(256,256),0,0,interpolation= cv2.INTER_NEAREST))'''"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dUEdcAfXNoiG",
        "outputId": "7a1c6fe9-9788-4387-ed5c-7fa9c3b40da6"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Thu Sep 23 10:32:20 2021       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 470.63.01    Driver Version: 460.32.03    CUDA Version: 11.2     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla P100-PCIE...  Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   34C    P0    28W / 250W |      0MiB / 16280MiB |      0%      Default |\n",
            "|                               |                      |                  N/A |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ]
        }
      ],
      "source": [
        "!nvidia-smi"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "8Nnr-O2hRkpc"
      },
      "outputs": [],
      "source": [
        "number=3"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 161
        },
        "id": "ZICr4kIhqQwZ",
        "outputId": "b3fb4dd9-7f1b-4c81-9ef5-e19cbfd8b602"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1080x1080 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "from matplotlib import pyplot as plt\n",
        "plt.figure(figsize=(15,15))\n",
        "plt.subplot(231)\n",
        "plt.imshow(mySet.trainingArrayX[number])\n",
        "plt.subplot(232)\n",
        "plt.imshow(mySet.trainingArrayY[number])\n",
        "number+=1"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dE12IZmmmCJT",
        "outputId": "fb449cb2-3f23-47a9-dec4-7e117181f02d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Collecting python-gdcm\n",
            "  Downloading python_gdcm-3.0.9.1-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (9.4 MB)\n",
            "\u001b[K     |████████████████████████████████| 9.4 MB 13.9 MB/s \n",
            "\u001b[?25hInstalling collected packages: python-gdcm\n",
            "Successfully installed python-gdcm-3.0.9.1\n"
          ]
        }
      ],
      "source": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2aavkH90BxWr",
        "outputId": "5bf9e871-2251-4b32-c2c5-21f10c80ac08"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "((8840, 256, 256), (8840, 256, 256, 3), (441, 256, 256), (441, 256, 256, 3))"
            ]
          },
          "execution_count": 7,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "mySet.trainingArrayX.shape,mySet.trainingArrayY.shape,mySet.testArrayX.shape,mySet.testArrayY.shape"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "EgQQ9qL2CUG5"
      },
      "outputs": [],
      "source": [
        "from skimage.transform import resize\n",
        "\n",
        "trainlistx,trainlisty=[],[]\n",
        "for i in range(len(mySet.trainingArrayX)):\n",
        "  trainlistx.append(resize(mySet.trainingArrayX[i],(256,256)))\n",
        "  trainlisty.append(resize(mySet.trainingArrayY[i],(256,256)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Kshoj-E63GPd"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "trainingArrayX=np.array(trainlistx)\n",
        "del trainlistx\n",
        "trainingArrayY=np.array(trainlisty)\n",
        "del trainlisty"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "j7uJ8CLU3t_H"
      },
      "outputs": [],
      "source": [
        "testlistx,testlisty=[],[]\n",
        "for i in range(len(mySet.testArrayX)):\n",
        "  testlistx.append(resize(mySet.testArrayX[i],(256,256)))\n",
        "  testlisty.append(resize(mySet.testArrayY[i],(256,256)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "gOmG5vDD4GDr"
      },
      "outputs": [],
      "source": [
        "testArrayX=np.array(testlistx)\n",
        "del testlistx\n",
        "testArrayY=np.array(testlisty)\n",
        "del testlisty"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 164
        },
        "id": "DK8VPA8I4MqB",
        "outputId": "b6b54848-cc7e-4118-e69f-703ddd8b0fb6"
      },
      "outputs": [],
      "source": [
        "trainingArrayX.shape,trainingArrayY.shape,testArrayX.shape,testArrayY.shape"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "dQdLFlhes3vO"
      },
      "outputs": [],
      "source": [
        "del mySet"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ilfZE3KIseDI"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "smooth=100\n",
        "# loss function and metrics\n",
        "def dice_coef(y_pred, Y):\n",
        "    y_flatten = K.flatten(Y)\n",
        "    y_pred_flatten = K.flatten(y_pred)\n",
        "    intersection = K.sum(y_flatten * y_pred_flatten)\n",
        "    dice = (0.2 * intersection + 1.0) / (K.sum(y_flatten) + K.sum(y_pred_flatten) + 1.0)\n",
        "    return dice\n",
        "\n",
        "def jacard_coef(y_pred, Y):\n",
        "    y_flatten = K.cast(K.flatten(Y[0,:,:,0:2]),tf.int8)\n",
        "    y_pred_flatten = K.cast(K.flatten(y_pred[0,:,:,0:2]),tf.int8)\n",
        "    intersection = K.sum(y_flatten * y_pred_flatten)\n",
        "    y_flatten_bool=K.cast(y_flatten,tf.bool)\n",
        "    y_pred_flatten_bool=K.cast(y_pred_flatten,tf.bool)\n",
        "    xor=K.sum(K.cast(tf.math.logical_xor(y_flatten_bool,y_pred_flatten_bool),tf.int8))\n",
        "    iou=intersection/(intersection+xor)\n",
        "    return iou\n",
        "\n",
        "def jacard_coef_loss(y_pred, Y):\n",
        "    return 1-jacard_coef(y_pred, Y)\n",
        "\n",
        "\n",
        "def dice_coef_loss(y_pred, Y):\n",
        "    return -dice_coef(y_pred, Y)\n",
        "  \n",
        "def iou(y_pred, y_true):\n",
        "\n",
        "    y_pred=K.flatten(y_pred)\n",
        "    y_true=K.flatten(y_true)\n",
        "    intersection=tf.keras.backend.any(tf.keras.backend.stack([y_pred, y_true], axis=0), axis=0)\n",
        "    union=tf.keras.backend.all(tf.keras.backend.stack([y_pred, y_true], axis=0), axis=0)\n",
        "    intersection = K.cast(intersection,\"int32\")\n",
        "    union=K.cast(union,\"int32\")\n",
        "    '''intersection=K.logical_and(y_pred[0:2],y_true[0:2])\n",
        "    union=K.logical_or(y_pred[0:2],y_true[0:2])'''\n",
        "    iou=K.sum(intersection)/K.sum(union)\n",
        "    return iou\n",
        "  \n",
        "\n",
        "def iou_coef(y_true, y_pred, smooth=1):\n",
        "  intersection = K.sum(K.abs(y_true * y_pred), axis=[1,2])\n",
        "  union = K.sum(y_true,[1,2])+K.sum(y_pred,[1,2])-intersection\n",
        "  iou = K.mean((intersection + smooth) / (union + smooth), axis=0)\n",
        "  return iou\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "w_8yzP0bqDSF"
      },
      "outputs": [],
      "source": [
        "def mean_iou(y_pred, y_true):\n",
        "  y_pred=K.flatten(y_pred)\n",
        "  y_true=K.flatten(y_true)\n",
        "  diffOne=tf.math.not_equal(y_pred[0,:,:,0],0)\n",
        "  diffTwo=tf.math.not_equal(y_true[0,:,:,0],0)\n",
        "  diff=K.cast(tf.math.equal(diffOne,diffTwo),tf.int8)\n",
        "  neg=K.cast(tf.math.not_equal(y_pred[0,:,:,0],y_true[0,:,:,0]),tf.int8)\n",
        "\n",
        "  diffOneTwo=tf.math.not_equal(y_pred[0,:,:,1],0)\n",
        "  diffTwoTwo=tf.math.not_equal(y_true[0,:,:,1],0)\n",
        "  diffTwo=K.cast(tf.math.equal(diffOneTwo,diffTwoTwo),tf.int8)\n",
        "  negTwo=K.cast(tf.math.not_equal(y_pred[0,:,:,1],y_true[0,:,:,1]),tf.int8)\n",
        "\n",
        "  int1=K.sum(diff)/(K.sum(diff)+K.sum(neg))\n",
        "  int2=K.sum(diffTwo)/(K.sum(diffTwo)+K.sum(negTwo))\n",
        "  print(int1,int2)\n",
        "  return (int1+int2)/2\n",
        "  "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "KLSHhr3JkgTj"
      },
      "outputs": [],
      "source": [
        "import attenunet"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IbmkpU_AF1fb",
        "outputId": "61179ca4-38bf-4435-aa8b-5b42610fdfe3"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"Attention_UNet\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            [(None, 256, 256, 3) 0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv2d (Conv2D)                 (None, 256, 256, 64) 1792        input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization (BatchNorma (None, 256, 256, 64) 256         conv2d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "activation (Activation)         (None, 256, 256, 64) 0           batch_normalization[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_1 (Conv2D)               (None, 256, 256, 64) 36928       activation[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_1 (BatchNor (None, 256, 256, 64) 256         conv2d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_1 (Activation)       (None, 256, 256, 64) 0           batch_normalization_1[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D)    (None, 128, 128, 64) 0           activation_1[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_2 (Conv2D)               (None, 128, 128, 128 73856       max_pooling2d[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_2 (BatchNor (None, 128, 128, 128 512         conv2d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_2 (Activation)       (None, 128, 128, 128 0           batch_normalization_2[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_3 (Conv2D)               (None, 128, 128, 128 147584      activation_2[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_3 (BatchNor (None, 128, 128, 128 512         conv2d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_3 (Activation)       (None, 128, 128, 128 0           batch_normalization_3[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2D)  (None, 64, 64, 128)  0           activation_3[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_4 (Conv2D)               (None, 64, 64, 256)  295168      max_pooling2d_1[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_4 (BatchNor (None, 64, 64, 256)  1024        conv2d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_4 (Activation)       (None, 64, 64, 256)  0           batch_normalization_4[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_5 (Conv2D)               (None, 64, 64, 256)  590080      activation_4[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_5 (BatchNor (None, 64, 64, 256)  1024        conv2d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_5 (Activation)       (None, 64, 64, 256)  0           batch_normalization_5[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2D)  (None, 32, 32, 256)  0           activation_5[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_6 (Conv2D)               (None, 32, 32, 512)  1180160     max_pooling2d_2[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_6 (BatchNor (None, 32, 32, 512)  2048        conv2d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_6 (Activation)       (None, 32, 32, 512)  0           batch_normalization_6[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_7 (Conv2D)               (None, 32, 32, 512)  2359808     activation_6[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_7 (BatchNor (None, 32, 32, 512)  2048        conv2d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_7 (Activation)       (None, 32, 32, 512)  0           batch_normalization_7[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2D)  (None, 16, 16, 512)  0           activation_7[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_8 (Conv2D)               (None, 16, 16, 1024) 4719616     max_pooling2d_3[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_8 (BatchNor (None, 16, 16, 1024) 4096        conv2d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_8 (Activation)       (None, 16, 16, 1024) 0           batch_normalization_8[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_9 (Conv2D)               (None, 16, 16, 1024) 9438208     activation_8[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_9 (BatchNor (None, 16, 16, 1024) 4096        conv2d_9[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_9 (Activation)       (None, 16, 16, 1024) 0           batch_normalization_9[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_10 (Conv2D)              (None, 16, 16, 512)  524800      activation_9[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_10 (BatchNo (None, 16, 16, 512)  2048        conv2d_10[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_10 (Activation)      (None, 16, 16, 512)  0           batch_normalization_10[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_12 (Conv2D)              (None, 16, 16, 512)  262656      activation_10[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_transpose (Conv2DTranspo (None, 16, 16, 512)  2359808     conv2d_12[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_11 (Conv2D)              (None, 16, 16, 512)  1049088     activation_7[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "add (Add)                       (None, 16, 16, 512)  0           conv2d_transpose[0][0]           \n",
            "                                                                 conv2d_11[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_11 (Activation)      (None, 16, 16, 512)  0           add[0][0]                        \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_13 (Conv2D)              (None, 16, 16, 1)    513         activation_11[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_12 (Activation)      (None, 16, 16, 1)    0           conv2d_13[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d (UpSampling2D)    (None, 32, 32, 1)    0           activation_12[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "lambda (Lambda)                 (None, 32, 32, 512)  0           up_sampling2d[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "multiply (Multiply)             (None, 32, 32, 512)  0           lambda[0][0]                     \n",
            "                                                                 activation_7[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_14 (Conv2D)              (None, 32, 32, 512)  262656      multiply[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d_1 (UpSampling2D)  (None, 32, 32, 1024) 0           activation_9[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_11 (BatchNo (None, 32, 32, 512)  2048        conv2d_14[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate (Concatenate)       (None, 32, 32, 1536) 0           up_sampling2d_1[0][0]            \n",
            "                                                                 batch_normalization_11[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_15 (Conv2D)              (None, 32, 32, 512)  7078400     concatenate[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_12 (BatchNo (None, 32, 32, 512)  2048        conv2d_15[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_13 (Activation)      (None, 32, 32, 512)  0           batch_normalization_12[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_16 (Conv2D)              (None, 32, 32, 512)  2359808     activation_13[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_13 (BatchNo (None, 32, 32, 512)  2048        conv2d_16[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_14 (Activation)      (None, 32, 32, 512)  0           batch_normalization_13[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_17 (Conv2D)              (None, 32, 32, 256)  131328      activation_14[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_14 (BatchNo (None, 32, 32, 256)  1024        conv2d_17[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_15 (Activation)      (None, 32, 32, 256)  0           batch_normalization_14[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_19 (Conv2D)              (None, 32, 32, 256)  65792       activation_15[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_transpose_1 (Conv2DTrans (None, 32, 32, 256)  590080      conv2d_19[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_18 (Conv2D)              (None, 32, 32, 256)  262400      activation_5[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "add_1 (Add)                     (None, 32, 32, 256)  0           conv2d_transpose_1[0][0]         \n",
            "                                                                 conv2d_18[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_16 (Activation)      (None, 32, 32, 256)  0           add_1[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_20 (Conv2D)              (None, 32, 32, 1)    257         activation_16[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_17 (Activation)      (None, 32, 32, 1)    0           conv2d_20[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d_2 (UpSampling2D)  (None, 64, 64, 1)    0           activation_17[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "lambda_1 (Lambda)               (None, 64, 64, 256)  0           up_sampling2d_2[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "multiply_1 (Multiply)           (None, 64, 64, 256)  0           lambda_1[0][0]                   \n",
            "                                                                 activation_5[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_21 (Conv2D)              (None, 64, 64, 256)  65792       multiply_1[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d_3 (UpSampling2D)  (None, 64, 64, 512)  0           activation_14[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_15 (BatchNo (None, 64, 64, 256)  1024        conv2d_21[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_1 (Concatenate)     (None, 64, 64, 768)  0           up_sampling2d_3[0][0]            \n",
            "                                                                 batch_normalization_15[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_22 (Conv2D)              (None, 64, 64, 256)  1769728     concatenate_1[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_16 (BatchNo (None, 64, 64, 256)  1024        conv2d_22[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_18 (Activation)      (None, 64, 64, 256)  0           batch_normalization_16[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_23 (Conv2D)              (None, 64, 64, 256)  590080      activation_18[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_17 (BatchNo (None, 64, 64, 256)  1024        conv2d_23[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_19 (Activation)      (None, 64, 64, 256)  0           batch_normalization_17[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_24 (Conv2D)              (None, 64, 64, 128)  32896       activation_19[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_18 (BatchNo (None, 64, 64, 128)  512         conv2d_24[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_20 (Activation)      (None, 64, 64, 128)  0           batch_normalization_18[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_26 (Conv2D)              (None, 64, 64, 128)  16512       activation_20[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_transpose_2 (Conv2DTrans (None, 64, 64, 128)  147584      conv2d_26[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_25 (Conv2D)              (None, 64, 64, 128)  65664       activation_3[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "add_2 (Add)                     (None, 64, 64, 128)  0           conv2d_transpose_2[0][0]         \n",
            "                                                                 conv2d_25[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_21 (Activation)      (None, 64, 64, 128)  0           add_2[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_27 (Conv2D)              (None, 64, 64, 1)    129         activation_21[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_22 (Activation)      (None, 64, 64, 1)    0           conv2d_27[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d_4 (UpSampling2D)  (None, 128, 128, 1)  0           activation_22[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "lambda_2 (Lambda)               (None, 128, 128, 128 0           up_sampling2d_4[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "multiply_2 (Multiply)           (None, 128, 128, 128 0           lambda_2[0][0]                   \n",
            "                                                                 activation_3[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_28 (Conv2D)              (None, 128, 128, 128 16512       multiply_2[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d_5 (UpSampling2D)  (None, 128, 128, 256 0           activation_19[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_19 (BatchNo (None, 128, 128, 128 512         conv2d_28[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_2 (Concatenate)     (None, 128, 128, 384 0           up_sampling2d_5[0][0]            \n",
            "                                                                 batch_normalization_19[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_29 (Conv2D)              (None, 128, 128, 128 442496      concatenate_2[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_20 (BatchNo (None, 128, 128, 128 512         conv2d_29[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_23 (Activation)      (None, 128, 128, 128 0           batch_normalization_20[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_30 (Conv2D)              (None, 128, 128, 128 147584      activation_23[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_21 (BatchNo (None, 128, 128, 128 512         conv2d_30[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_24 (Activation)      (None, 128, 128, 128 0           batch_normalization_21[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_31 (Conv2D)              (None, 128, 128, 64) 8256        activation_24[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_22 (BatchNo (None, 128, 128, 64) 256         conv2d_31[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_25 (Activation)      (None, 128, 128, 64) 0           batch_normalization_22[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_33 (Conv2D)              (None, 128, 128, 64) 4160        activation_25[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_transpose_3 (Conv2DTrans (None, 128, 128, 64) 36928       conv2d_33[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_32 (Conv2D)              (None, 128, 128, 64) 16448       activation_1[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "add_3 (Add)                     (None, 128, 128, 64) 0           conv2d_transpose_3[0][0]         \n",
            "                                                                 conv2d_32[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_26 (Activation)      (None, 128, 128, 64) 0           add_3[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_34 (Conv2D)              (None, 128, 128, 1)  65          activation_26[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_27 (Activation)      (None, 128, 128, 1)  0           conv2d_34[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d_6 (UpSampling2D)  (None, 256, 256, 1)  0           activation_27[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "lambda_3 (Lambda)               (None, 256, 256, 64) 0           up_sampling2d_6[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "multiply_3 (Multiply)           (None, 256, 256, 64) 0           lambda_3[0][0]                   \n",
            "                                                                 activation_1[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_35 (Conv2D)              (None, 256, 256, 64) 4160        multiply_3[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "up_sampling2d_7 (UpSampling2D)  (None, 256, 256, 128 0           activation_24[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_23 (BatchNo (None, 256, 256, 64) 256         conv2d_35[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_3 (Concatenate)     (None, 256, 256, 192 0           up_sampling2d_7[0][0]            \n",
            "                                                                 batch_normalization_23[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_36 (Conv2D)              (None, 256, 256, 64) 110656      concatenate_3[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_24 (BatchNo (None, 256, 256, 64) 256         conv2d_36[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_28 (Activation)      (None, 256, 256, 64) 0           batch_normalization_24[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_37 (Conv2D)              (None, 256, 256, 64) 36928       activation_28[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_25 (BatchNo (None, 256, 256, 64) 256         conv2d_37[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_29 (Activation)      (None, 256, 256, 64) 0           batch_normalization_25[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_38 (Conv2D)              (None, 256, 256, 3)  195         activation_29[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_26 (BatchNo (None, 256, 256, 3)  12          conv2d_38[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_30 (Activation)      (None, 256, 256, 3)  0           batch_normalization_26[0][0]     \n",
            "==================================================================================================\n",
            "Total params: 37,334,803\n",
            "Trainable params: 37,319,181\n",
            "Non-trainable params: 15,622\n",
            "__________________________________________________________________________________________________\n",
            "None\n"
          ]
        }
      ],
      "source": [
        "unet_model = attenunet.Attention_UNet((256,256,3),num_classes=3,dropout=0.2)\n",
        "att_unet_model = unet_model.build_attention_unit()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "_pjw06pR8hP7"
      },
      "outputs": [],
      "source": [
        "tf.keras.utils.plot_model(att_unet_model)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Ey3pHPrR8-Lf"
      },
      "outputs": [],
      "source": [
        "att_unet_model.load_weights(\"/content/drive/MyDrive/SegmentationNewModels/HighestmodelFifth.h5\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ywbOpLrBtZuT"
      },
      "outputs": [],
      "source": [
        "focal_loss = sm.losses.CategoricalCELoss()\n",
        "\n",
        "adam=tf.keras.optimizers.Adam()\n",
        "metric = [sm.metrics.IOUScore(threshold=0.5), sm.metrics.FScore(threshold=0.5)]\n",
        "att_unet_model.compile(loss=focal_loss, optimizer=adam, \n",
        "              metrics=metric)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "dICu_HJIukPV"
      },
      "outputs": [],
      "source": [
        "checkpoint_filepath=\"/content/drive/MyDrive/attUnetModels/mehakif.h5\"\n",
        "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
        "    filepath=checkpoint_filepath,\n",
        "    save_weights_only=True,\n",
        "    monitor='val_iou_score',\n",
        "    mode='max',\n",
        "    save_best_only=True)\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 463
        },
        "id": "4M4tqiC7Ltqw",
        "outputId": "54245b96-590f-4a43-a2c1-75ab650b154c"
      },
      "outputs": [],
      "source": [
        "history = att_unet_model.fit(mySet.trainingArrayX, mySet.trainingArrayY, \n",
        "                             verbose=1,\n",
        "                             batch_size = 8,\n",
        "                             validation_data=(mySet.testArrayX,mySet.testArrayY),\n",
        "                             shuffle=False,\n",
        "                             epochs=25,\n",
        "                             callbacks=[model_checkpoint_callback])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IUgEmrwbF5Is",
        "outputId": "42f9f346-0acd-4123-f213-515371f68f0b"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "14/14 [==============================] - 4s 207ms/step - loss: 0.0590 - dice_coef: 0.9904 - jacard_coef: -0.1820\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "[0.05901692807674408, 0.9904336333274841, -0.18195804953575134]"
            ]
          },
          "execution_count": 252,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "att_unet_model.evaluate(mySet.testArrayX,mySet.testArrayY,verbose=1)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "lUvm81T72Q1Z"
      },
      "outputs": [],
      "source": [
        "num_img_test = 0"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ln-gcOCIKbLD"
      },
      "outputs": [],
      "source": [
        "from skimage.transform import resize\n",
        "\n",
        "img=cv2.imread(\"/content/drive/MyDrive/NewDatasetsBG/Bone/TestSet/No/10100.png\")\n",
        "img=img.astype(np.float32)\n",
        "img/=255.\n",
        "img=resize(img,(256,256))\n",
        "img=img.reshape(1,256,256,3)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "uuy2l0MiKud8"
      },
      "outputs": [],
      "source": [
        "att_unet_model1 = unet_model.build_attention_unit()\n",
        "att_unet_model1.load_weights(\"/content/drive/MyDrive/SegmentationNewModels/HighestmodelOne.h5\")\n",
        "att_unet_model2 = unet_model.build_attention_unit()\n",
        "att_unet_model2.load_weights(\"/content/drive/MyDrive/SegmentationNewModels/HighestmodelTwo.h5\")\n",
        "att_unet_model3 = unet_model.build_attention_unit()\n",
        "att_unet_model3.load_weights(\"/content/drive/MyDrive/SegmentationNewModels/HighestmodelThree.h5\")\n",
        "att_unet_model4 = unet_model.build_attention_unit()\n",
        "att_unet_model4.load_weights(\"/content/drive/MyDrive/SegmentationNewModels/HighestmodelFour.h5\")\n",
        "att_unet_model5 = unet_model.build_attention_unit()\n",
        "att_unet_model5.load_weights(\"/content/drive/MyDrive/SegmentationNewModels/HighestmodelFifth.h5\")\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "vWw3ZYqoaZA8"
      },
      "outputs": [],
      "source": [
        "num_img_test=0"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 310
        },
        "id": "Ts_QkpdVt6vU",
        "outputId": "2df92cce-122f-497a-da8a-a0f3f801c47e"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1080x720 with 3 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "\n",
        "img_test = cv2.imread(\"/content/drive/MyDrive/Segment/BoneNewTwo/TestSet/Annotation/\"+mySet.nameArray[num_img_test])\n",
        "test_label = cv2.resize(cv2.imread(\"/content/testFolder/\"+mySet.nameArray[num_img_test]),(512,512),interpolation=cv2.INTER_CUBIC)\n",
        "\n",
        "img_test_input = np.expand_dims(img_test, 0)\n",
        "\n",
        "#pred = ((model.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "'''pred2 = ((att_unet_model2.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred3 = ((att_unet_model3.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred4 = ((att_unet_model4.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred5 = ((att_unet_model5.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)'''\n",
        "#pred = (att_unet_model.predict(img_test_input)[0,:,:,0] > 0.5).astype(np.uint8)\n",
        "#pred2 = (att_unet_model.predict(img_test_input)[0,:,:,1] > 0.5).astype(np.uint8)\n",
        "# Visualize the result \n",
        "plt.figure(figsize=(15, 10))\n",
        "plt.subplot(231)\n",
        "plt.title('Testing Image')\n",
        "plt.imshow(img_test, cmap='gray')\n",
        "plt.subplot(232)\n",
        "plt.title('Testing Label')\n",
        "plt.imshow(test_label, cmap='gray')\n",
        "plt.subplot(233)\n",
        "plt.title('Prediction on test image')\n",
        "plt.imshow(test_label, cmap='gray')\n",
        "\n",
        "\n",
        "num_img_test+=1\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 181
        },
        "id": "GGF0uUHs-xLi",
        "outputId": "3a9cc892-fd56-4aea-c200-3b6546576589"
      },
      "outputs": [
        {
          "ename": "NameError",
          "evalue": "ignored",
          "output_type": "error",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-35-e0173067867e>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mmatplotlib\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mpyplot\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mplt\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mimsave\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"perd4.png\"\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mcv2\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcvtColor\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpred4\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mcv2\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mCOLOR_BGR2RGB\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m: name 'pred4' is not defined"
          ]
        }
      ],
      "source": [
        "from matplotlib import pyplot as plt\n",
        "plt.imsave(\"perd4.png\",cv2.cvtColor(pred4,cv2.COLOR_BGR2RGB))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Kcf_Gx3O_q-g"
      },
      "outputs": [],
      "source": [
        "import cv2"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "AysIO9MfQ5hk"
      },
      "outputs": [],
      "source": [
        "for i in range(len(mySet.nameArray)):\n",
        "  img_test = mySet.testArrayX[i]\n",
        "  img_test_input = np.expand_dims(img_test, 0)\n",
        "  plt.imsave(\"/content/testFolder/\"+str(mySet.nameArray[i]),cv2.cvtColor(((model.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8),cv2.COLOR_BGR2RGB))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "xrTDHAm8BK57"
      },
      "outputs": [],
      "source": [
        "!zip -r \"/content/testFolder.zip\" \"/content/testFolder\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Gv1JwrL7S6v_"
      },
      "outputs": [],
      "source": [
        "target=res\n",
        "prediction=resized\n",
        "intersection = np.logical_and(target, prediction)\n",
        "union = np.logical_or(target, prediction)\n",
        "iou_score = np.sum(intersection) / np.sum(union)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "HTc_rqfIVkQU"
      },
      "outputs": [],
      "source": [
        "iou_score"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7KI7uKIi-y52"
      },
      "source": [
        "# Model 2"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "-8l_G8iC-1HM"
      },
      "outputs": [],
      "source": [
        "def multi_unet_model(n_classes=4, IMG_HEIGHT=256, IMG_WIDTH=256, IMG_CHANNELS=1):\n",
        "#Build the model\n",
        "    inputs = Input((IMG_HEIGHT, IMG_WIDTH, IMG_CHANNELS))\n",
        "    #s = Lambda(lambda x: x / 255)(inputs)   #No need for this if we normalize our inputs beforehand\n",
        "    s = inputs\n",
        "\n",
        "    #Contraction path\n",
        "    c1 = Conv2D(16, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(s)\n",
        "    c1 = Dropout(0.2)(c1)  # Original 0.1\n",
        "    c1 = Conv2D(16, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c1)\n",
        "    p1 = MaxPooling2D((2, 2))(c1)\n",
        "    \n",
        "    c2 = Conv2D(32, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(p1)\n",
        "    c2 = Dropout(0.2)(c2)  # Original 0.1\n",
        "    c2 = Conv2D(32, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c2)\n",
        "    p2 = MaxPooling2D((2, 2))(c2)\n",
        "     \n",
        "    c3 = Conv2D(64, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(p2)\n",
        "    c3 = Dropout(0.2)(c3)\n",
        "    c3 = Conv2D(64, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c3)\n",
        "    p3 = MaxPooling2D((2, 2))(c3)\n",
        "     \n",
        "    c4 = Conv2D(128, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(p3)\n",
        "    c4 = Dropout(0.2)(c4)\n",
        "    c4 = Conv2D(128, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c4)\n",
        "    p4 = MaxPooling2D(pool_size=(2, 2))(c4)\n",
        "     \n",
        "    c5 = Conv2D(256, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(p4)\n",
        "    c5 = Dropout(0.3)(c5)\n",
        "    c5 = Conv2D(256, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c5)\n",
        "    \n",
        "    #Expansive path \n",
        "    u6 = Conv2DTranspose(128, (2, 2), strides=(2, 2), padding='same')(c5)\n",
        "    u6 = concatenate([u6, c4])\n",
        "    c6 = Conv2D(128, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(u6)\n",
        "    c6 = Dropout(0.2)(c6)\n",
        "    c6 = Conv2D(128, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c6)\n",
        "     \n",
        "    u7 = Conv2DTranspose(64, (2, 2), strides=(2, 2), padding='same')(c6)\n",
        "    u7 = concatenate([u7, c3])\n",
        "    c7 = Conv2D(64, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(u7)\n",
        "    c7 = Dropout(0.2)(c7)\n",
        "    c7 = Conv2D(64, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c7)\n",
        "     \n",
        "    u8 = Conv2DTranspose(32, (2, 2), strides=(2, 2), padding='same')(c7)\n",
        "    u8 = concatenate([u8, c2])\n",
        "    c8 = Conv2D(32, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(u8)\n",
        "    c8 = Dropout(0.2)(c8)  # Original 0.1\n",
        "    c8 = Conv2D(32, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c8)\n",
        "     \n",
        "    u9 = Conv2DTranspose(16, (2, 2), strides=(2, 2), padding='same')(c8)\n",
        "    u9 = concatenate([u9, c1], axis=3)\n",
        "    c9 = Conv2D(16, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(u9)\n",
        "    c9 = Dropout(0.2)(c9)  # Original 0.1\n",
        "    c9 = Conv2D(16, (3, 3), activation='relu', kernel_initializer='he_normal', padding='same')(c9)\n",
        "     \n",
        "    outputs = Conv2D(n_classes, (1, 1), activation='softmax')(c9)\n",
        "     \n",
        "    model = Model(inputs=[inputs], outputs=[outputs])\n",
        "    \n",
        "    #NOTE: Compile the model in the main program to make it easy to test with various loss functions\n",
        "    #model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])\n",
        "    \n",
        "    #model.summary()\n",
        "    \n",
        "    return model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "SErcRHz8AcRc"
      },
      "outputs": [],
      "source": [
        "multiunet=multi_unet_model(3,512,512)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "c6CYbZQ51AkR"
      },
      "outputs": [],
      "source": [
        "newmulti=multi_unet_model(3,512,512,3)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "rDwjbYJ_aNLy"
      },
      "outputs": [],
      "source": [
        "newmulti.load_weights(\"/content/drive/MyDrive/sotaSegMod/model2w.h5\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Un43beBCdhMH"
      },
      "outputs": [],
      "source": [
        "multiunet.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "sXgnp_8rVkYF"
      },
      "outputs": [],
      "source": [
        "multiunet.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy',jacard_coef])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "UttXtYXWVwtX"
      },
      "outputs": [],
      "source": [
        "checkpoint_filepath=\"/content/drive/MyDrive/sotaSegMod/modelgrayscaleNEWMODEL2.h5\"\n",
        "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
        "    filepath=checkpoint_filepath,\n",
        "    monitor='val_jacard_coef',\n",
        "    mode='max',\n",
        "    save_weights_only=True,\n",
        "    save_best_only=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "HhCBAl2aV9CH"
      },
      "outputs": [],
      "source": [
        "history = multiunet.fit(mySet.trainingArrayX, mySet.trainingArrayY, \n",
        "                             verbose=1,\n",
        "                             batch_size = 16,\n",
        "                             validation_data=(mySet.testArrayX,mySet.testArrayY),\n",
        "                             shuffle=False,\n",
        "                             epochs=200,\n",
        "                             callbacks=[model_checkpoint_callback])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "LvnIk-blW1et"
      },
      "outputs": [],
      "source": [
        "multiunet.evaluate(mySet.testArrayX,mySet.testArrayY)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "zXhjTkiECTbr"
      },
      "outputs": [],
      "source": [
        "for i in range(len(mySet.nameArray)):\n",
        "  img_test = mySet.testArrayX[i]\n",
        "  img_test_input = np.expand_dims(img_test, 0)\n",
        "  plt.imsave(\"/content/testFolder/\"+str(mySet.nameArray[i]),(((model.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "VzNCxi9U9MaA"
      },
      "outputs": [],
      "source": [
        "num_img_test=0"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 310
        },
        "id": "jMn0B3KQWBtP",
        "outputId": "c1072081-75f1-410a-ea9b-55989f3a8300"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1080x720 with 3 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "\n",
        "img_test = mySet.testArrayX[num_img_test]\n",
        "test_label = mySet.testArrayY[num_img_test]\n",
        "\n",
        "img_test_input = np.expand_dims(img_test, 0)\n",
        "\n",
        "pred = ((model.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "'''pred2 = ((att_unet_model2.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred3 = ((att_unet_model3.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred4 = ((att_unet_model4.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred5 = ((att_unet_model5.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)'''\n",
        "#pred = (att_unet_model.predict(img_test_input)[0,:,:,0] > 0.5).astype(np.uint8)\n",
        "#pred2 = (att_unet_model.predict(img_test_input)[0,:,:,1] > 0.5).astype(np.uint8)\n",
        "# Visualize the result \n",
        "plt.figure(figsize=(15, 10))\n",
        "plt.subplot(231)\n",
        "plt.title('Testing Image')\n",
        "plt.imshow(img_test, cmap='gray')\n",
        "plt.subplot(232)\n",
        "plt.title('Testing Label')\n",
        "plt.imshow(test_label[:,:,:], cmap='gray')\n",
        "plt.subplot(233)\n",
        "plt.title('Prediction on test image')\n",
        "plt.imshow(pred, cmap='gray')\n",
        "\n",
        "\n",
        "num_img_test+=1\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Oct2nodPXBlp"
      },
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NsjnuZ1deiEV"
      },
      "source": [
        "# MODEL 3\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "EByqsyhZe6B6"
      },
      "outputs": [],
      "source": [
        "IMAGE_SIZE = 512\n",
        "EPOCHS = 30\n",
        "BATCH = 8\n",
        "LR = 1e-4\n",
        "\n",
        "PATH = \"/content/\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "-ECWSt-Aey-Y"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import numpy as np\n",
        "import cv2\n",
        "from glob import glob\n",
        "import tensorflow as tf\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "from tensorflow.keras.layers import Conv2D, Activation, BatchNormalization\n",
        "from tensorflow.keras.layers import UpSampling2D, Input, Concatenate\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.applications import MobileNetV2\n",
        "from tensorflow.keras.callbacks import EarlyStopping, ReduceLROnPlateau\n",
        "from tensorflow.keras.metrics import Recall, Precision\n",
        "from tensorflow.keras import backend as K"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "mBf8j_VkekKC"
      },
      "outputs": [],
      "source": [
        "def model():\n",
        "    inputs = Input(shape=(IMAGE_SIZE, IMAGE_SIZE, 3), name=\"input_image\")\n",
        "    \n",
        "    encoder = MobileNetV2(input_tensor=inputs, weights=\"imagenet\", include_top=False, alpha=0.35)\n",
        "    skip_connection_names = [\"input_image\", \"block_1_expand_relu\", \"block_3_expand_relu\", \"block_6_expand_relu\"]\n",
        "    encoder_output = encoder.get_layer(\"block_13_expand_relu\").output\n",
        "    \n",
        "    f = [16, 32, 48, 64]\n",
        "    x = encoder_output\n",
        "    for i in range(1, len(skip_connection_names)+1, 1):\n",
        "        x_skip = encoder.get_layer(skip_connection_names[-i]).output\n",
        "        x = UpSampling2D((2, 2))(x)\n",
        "        x = Concatenate()([x, x_skip])\n",
        "        \n",
        "        x = Conv2D(f[-i], (3, 3), padding=\"same\")(x)\n",
        "        x = BatchNormalization()(x)\n",
        "        x = Activation(\"relu\")(x)\n",
        "        \n",
        "        x = Conv2D(f[-i], (3, 3), padding=\"same\")(x)\n",
        "        x = BatchNormalization()(x)\n",
        "        x = Activation(\"relu\")(x)\n",
        "        \n",
        "    x = Conv2D(3, (1, 1), padding=\"same\")(x)\n",
        "    x = Activation(\"softmax\")(x)\n",
        "    \n",
        "    model = Model(inputs, x)\n",
        "    return model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "jA18VO7lel9Z"
      },
      "outputs": [],
      "source": [
        "modelim=model()\n",
        "modelim.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "kt7d2VGgenFb"
      },
      "outputs": [],
      "source": [
        "modelim.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy',jacard_coef])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "KSBE_c3FhtpH"
      },
      "outputs": [],
      "source": [
        "checkpoint_filepath=\"/content/drive/MyDrive/SegmentationNewModels/modelim22.h5\"\n",
        "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
        "    filepath=checkpoint_filepath,\n",
        "    save_weights_only=True,\n",
        "    monitor='val_jacard_coef',\n",
        "    mode='max',\n",
        "    save_best_only=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Fd129h4Vhu6Z"
      },
      "outputs": [],
      "source": [
        "history = modelim.fit(mySet.trainingArrayX, mySet.trainingArrayY, \n",
        "                             verbose=1,\n",
        "                             batch_size = 4,\n",
        "                             validation_data=(mySet.testArrayX,mySet.testArrayY),\n",
        "                             shuffle=False,\n",
        "                             epochs=50,\n",
        "                             callbacks=[model_checkpoint_callback])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ZfvIZDYohxya"
      },
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9BvpgZqIpT-W"
      },
      "source": [
        "# MODEL 4\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "vdPqPujApWyk"
      },
      "outputs": [],
      "source": [
        "from tensorflow.keras.layers import Conv2D, BatchNormalization, Activation, MaxPool2D, Conv2DTranspose, Concatenate, Input\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.applications import VGG16\n",
        "\n",
        "def conv_block(input, num_filters):\n",
        "    x = Conv2D(num_filters, 3, padding=\"same\")(input)\n",
        "    x = BatchNormalization()(x)\n",
        "    x = Activation(\"relu\")(x)\n",
        "\n",
        "    x = Conv2D(num_filters, 3, padding=\"same\")(x)\n",
        "    x = BatchNormalization()(x)\n",
        "    x = Activation(\"relu\")(x)\n",
        "\n",
        "    return x\n",
        "\n",
        "def decoder_block(input, skip_features, num_filters):\n",
        "    x = Conv2DTranspose(num_filters, (2, 2), strides=2, padding=\"same\")(input)\n",
        "    x = Concatenate()([x, skip_features])\n",
        "    x = conv_block(x, num_filters)\n",
        "    return x\n",
        "\n",
        "def build_vgg16_unet(input_shape):\n",
        "    \"\"\" Input \"\"\"\n",
        "    inputs = Input(input_shape)\n",
        "\n",
        "    \"\"\" Pre-trained VGG16 Model \"\"\"\n",
        "    vgg16 = VGG16(include_top=False, weights=\"imagenet\", input_tensor=inputs)\n",
        "\n",
        "    \"\"\" Encoder \"\"\"\n",
        "    s1 = vgg16.get_layer(\"block1_conv2\").output         ## (512 x 512)\n",
        "    s2 = vgg16.get_layer(\"block2_conv2\").output         ## (256 x 256)\n",
        "    s3 = vgg16.get_layer(\"block3_conv3\").output         ## (128 x 128)\n",
        "    s4 = vgg16.get_layer(\"block4_conv3\").output         ## (64 x 64)\n",
        "\n",
        "    \"\"\" Bridge \"\"\"\n",
        "    b1 = vgg16.get_layer(\"block5_conv3\").output         ## (32 x 32)\n",
        "\n",
        "    \"\"\" Decoder \"\"\"\n",
        "    d1 = decoder_block(b1, s4, 512)                     ## (64 x 64)\n",
        "    d2 = decoder_block(d1, s3, 256)                     ## (128 x 128)\n",
        "    d3 = decoder_block(d2, s2, 128)                     ## (256 x 256)\n",
        "    d4 = decoder_block(d3, s1, 64)                      ## (512 x 512)\n",
        "\n",
        "    \"\"\" Output \"\"\"\n",
        "    outputs = Conv2D(3, 1, padding=\"same\", activation=\"softmax\")(d4)\n",
        "\n",
        "    model = Model(inputs, outputs, name=\"VGG16_U-Net\")\n",
        "    return model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "QmJeirfdpcnB"
      },
      "outputs": [],
      "source": [
        "model = build_vgg16_unet((256,256,3))\n",
        "model.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Y-_26jzwpkJe"
      },
      "outputs": [],
      "source": [
        "model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy',jacard_coef])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "VFfoYxP3prwe"
      },
      "outputs": [],
      "source": [
        "checkpoint_filepath=\"/content/drive/MyDrive/SegmentationNewModels/vgg16unet.h5\"\n",
        "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
        "    filepath=checkpoint_filepath,\n",
        "    save_weights_only=True,\n",
        "    monitor='val_jacard_coef',\n",
        "    mode='max',\n",
        "    save_best_only=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "j6bjyHHpptxu"
      },
      "outputs": [],
      "source": [
        "history = model.fit(mySet.trainingArrayX, mySet.trainingArrayY, \n",
        "                             verbose=1,\n",
        "                             batch_size = 16,\n",
        "                             validation_data=(mySet.testArrayX,mySet.testArrayY),\n",
        "                             shuffle=False,\n",
        "                             epochs=50,\n",
        "                             callbacks=[model_checkpoint_callback])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "I6Gxt7X7EzYm"
      },
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LQ_hz2_y8ORn"
      },
      "source": [
        "# MODEL 5\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "mO4Rm26FG0e_"
      },
      "outputs": [],
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import os\n",
        "from tensorflow.keras.layers import *\n",
        "#from tensorflow.keras.layers import Conv2D, Conv2DTranspose\n",
        "from tensorflow.keras.models import *\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator, load_img\n",
        "from tensorflow.keras.preprocessing.image import img_to_array\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "wGl-Co6L8QdI"
      },
      "outputs": [],
      "source": [
        "input_height=256\n",
        "input_width=256\n",
        "n_classes=3\n",
        "VGG_Weights_path=\"/content/vgg16_weights_tf_dim_ordering_tf_kernels_notop.h5\"\n",
        "def FCN8(IMG_input, nClasses,  input_height=224, input_width=224):\n",
        "    ## input_height and width must be devisible by 32 because maxpooling with filter size = (2,2) is operated 5 times,\n",
        "    ## which makes the input_height and width 2^5 = 32 times smaller   \n",
        "    ## Block 1\n",
        "    x = Conv2D(64, (3, 3), activation='relu', padding='same', name='block1_conv1')(IMG_input)\n",
        "    x = Conv2D(64, (3, 3), activation='relu', padding='same', name='block1_conv2')(x)\n",
        "    x = MaxPooling2D((2, 2), strides=(2, 2), name='block1_pool')(x)\n",
        "    f1 = x\n",
        "    \n",
        "    # Block 2\n",
        "    x = Conv2D(128, (3, 3), activation='relu', padding='same', name='block2_conv1')(x)\n",
        "    x = Conv2D(128, (3, 3), activation='relu', padding='same', name='block2_conv2')(x)\n",
        "    x = MaxPooling2D((2, 2), strides=(2, 2), name='block2_pool')(x)\n",
        "    f2 = x\n",
        "\n",
        "    # Block 3\n",
        "    x = Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv1')(x)\n",
        "    x = Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv2')(x)\n",
        "    x = Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv3')(x)\n",
        "    x = MaxPooling2D((2, 2), strides=(2, 2), name='block3_pool')(x)\n",
        "    pool3 = x\n",
        "\n",
        "    # Block 4\n",
        "    x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv1')(x)\n",
        "    x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv2')(x)\n",
        "    x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv3')(x)\n",
        "    pool4 = MaxPooling2D((2, 2), strides=(2, 2), name='block4_pool')(x)## (None, 14, 14, 512) \n",
        "\n",
        "    # Block 5\n",
        "    x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv1')(pool4)\n",
        "    x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv2')(x)\n",
        "    x = Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv3')(x)\n",
        "    pool5 = MaxPooling2D((2, 2), strides=(2, 2), name='block5_pool')(x)## (None, 7, 7, 512)\n",
        "     \n",
        "    vgg  = Model(img_input, pool5)\n",
        "    vgg.load_weights(VGG_Weights_path) ## loading VGG weights for the encoder parts of FCN8\n",
        "    \n",
        "    o = (Conv2D(n_classes, (7, 7), activation='relu', padding='same', name=\"conv6\"))(pool5)\n",
        "        \n",
        "    ## 4 times upsamping for final layer\n",
        "    conv6_4 = Conv2DTranspose(n_classes, kernel_size=(4,4),  strides=(4,4), use_bias = False)(o)\n",
        "    ## 2 times upsampling for pool411\n",
        "    pool4_n = ( Conv2D(n_classes, (1, 1), activation='relu', padding='same', name=\"pool4_n\"))(pool4)\n",
        "    pool4_n_2 = (Conv2DTranspose(nClasses, kernel_size=(2,2), strides=(2,2), use_bias=False))(pool4_n)\n",
        "    #1 time for pool3\n",
        "    pool3_n = (Conv2D( nClasses , ( 1 , 1 ) , activation='relu' , padding='same', name=\"pool3_n\"))(pool3)\n",
        "        \n",
        "    o = Add(name=\"add\")([pool4_n_2, pool3_n, conv6_4 ])\n",
        "    o = Conv2DTranspose( nClasses , kernel_size=(8,8) ,  strides=(8,8) , use_bias=False)(o)\n",
        "    o = (Activation('softmax'))(o)\n",
        "    \n",
        "    model = Model(img_input, o)\n",
        "    return model\n",
        "\n",
        "img_input = Input(shape=(input_height,input_width, 3)) \n",
        "model = FCN8(img_input,n_classes,  \n",
        "             input_height = 224, \n",
        "             input_width = 224)\n",
        "model.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "83VnHdN8-MC5"
      },
      "outputs": [],
      "source": [
        "model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy',jacard_coef])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "primaR0z9Uo9"
      },
      "outputs": [],
      "source": [
        "checkpoint_filepath=\"/content/drive/MyDrive/SegmentationNewModels/fcnmodel.h5\"\n",
        "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
        "    filepath=checkpoint_filepath,\n",
        "    save_weights_only=True,\n",
        "    monitor='val_jacard_coef',\n",
        "    mode='max',\n",
        "    save_best_only=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "-BmEG7aT-Ab6"
      },
      "outputs": [],
      "source": [
        "history = model.fit(mySet.trainingArrayX, mySet.trainingArrayY, \n",
        "                             verbose=1,\n",
        "                             batch_size = 16,\n",
        "                             validation_data=(mySet.testArrayX,mySet.testArrayY),\n",
        "                             shuffle=False,\n",
        "                             epochs=50,\n",
        "                             callbacks=[model_checkpoint_callback])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "B3aU82P0o2Pd"
      },
      "source": [
        "# MODEL 6\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "NMxxdXpho7w_"
      },
      "outputs": [],
      "source": [
        "from tensorflow.keras import layers\n",
        "import keras\n",
        "\n",
        "\n",
        "def get_model(img_size, num_classes):\n",
        "    inputs = keras.Input(shape=img_size + (3,))\n",
        "\n",
        "    ### [First half of the network: downsampling inputs] ###\n",
        "\n",
        "    # Entry block\n",
        "    x = layers.Conv2D(32, 3, strides=2, padding=\"same\")(inputs)\n",
        "    x = layers.BatchNormalization()(x)\n",
        "    x = layers.Activation(\"relu\")(x)\n",
        "\n",
        "    previous_block_activation = x  # Set aside residual\n",
        "\n",
        "    # Blocks 1, 2, 3 are identical apart from the feature depth.\n",
        "    for filters in [64, 128, 256]:\n",
        "        x = layers.Activation(\"relu\")(x)\n",
        "        x = layers.SeparableConv2D(filters, 3, padding=\"same\")(x)\n",
        "        x = layers.BatchNormalization()(x)\n",
        "\n",
        "        x = layers.Activation(\"relu\")(x)\n",
        "        x = layers.SeparableConv2D(filters, 3, padding=\"same\")(x)\n",
        "        x = layers.BatchNormalization()(x)\n",
        "\n",
        "        x = layers.MaxPooling2D(3, strides=2, padding=\"same\")(x)\n",
        "\n",
        "        # Project residual\n",
        "        residual = layers.Conv2D(filters, 1, strides=2, padding=\"same\")(\n",
        "            previous_block_activation\n",
        "        )\n",
        "        x = layers.add([x, residual])  # Add back residual\n",
        "        previous_block_activation = x  # Set aside next residual\n",
        "\n",
        "    ### [Second half of the network: upsampling inputs] ###\n",
        "\n",
        "    for filters in [256, 128, 64, 32]:\n",
        "        x = layers.Activation(\"relu\")(x)\n",
        "        x = layers.Conv2DTranspose(filters, 3, padding=\"same\")(x)\n",
        "        x = layers.BatchNormalization()(x)\n",
        "\n",
        "        x = layers.Activation(\"relu\")(x)\n",
        "        x = layers.Conv2DTranspose(filters, 3, padding=\"same\")(x)\n",
        "        x = layers.BatchNormalization()(x)\n",
        "\n",
        "        x = layers.UpSampling2D(2)(x)\n",
        "\n",
        "        # Project residual\n",
        "        residual = layers.UpSampling2D(2)(previous_block_activation)\n",
        "        residual = layers.Conv2D(filters, 1, padding=\"same\")(residual)\n",
        "        x = layers.add([x, residual])  # Add back residual\n",
        "        previous_block_activation = x  # Set aside next residual\n",
        "\n",
        "    # Add a per-pixel classification layer\n",
        "    outputs = layers.Conv2D(num_classes, 3, activation=\"softmax\", padding=\"same\")(x)\n",
        "\n",
        "    # Define the model\n",
        "    model = keras.Model(inputs, outputs)\n",
        "    return model\n",
        "\n",
        "\n",
        "# Free up RAM in case the model definition cells were run multiple times\n",
        "keras.backend.clear_session()\n",
        "\n",
        "# Build model\n",
        "model = get_model((256,256), 3)\n",
        "model.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "O07dH167pzeM"
      },
      "outputs": [],
      "source": [
        "model.compile(optimizer=\"rmsprop\", loss=\"categorical_crossentropy\")\n",
        "\n",
        "callbacks = [\n",
        "    keras.callbacks.ModelCheckpoint(\"oxford_segmentation.h5\", save_best_only=True)\n",
        "]\n",
        "\n",
        "# Train the model, doing validation at the end of each epoch.\n",
        "epochs = 15\n",
        "model.fit(mySet.trainingArrayX,mySet.trainingArrayY, epochs=epochs, validation_data=(mySet.testArrayX,mySet.testArrayY), callbacks=callbacks)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "HYN6shxWrVNP"
      },
      "outputs": [],
      "source": [
        "num_img_test=0"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "bDqCLRyHqRJL"
      },
      "outputs": [],
      "source": [
        "\n",
        "img_test = mySet.testArrayX[num_img_test]\n",
        "test_label = mySet.testArrayY[num_img_test]\n",
        "\n",
        "img_test_input = np.expand_dims(img_test, 0)\n",
        "\n",
        "pred = ((att_unet_model.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "'''pred2 = ((att_unet_model2.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred3 = ((att_unet_model3.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred4 = ((att_unet_model4.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred5 = ((att_unet_model5.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)'''\n",
        "#pred = (att_unet_model.predict(img_test_input)[0,:,:,0] > 0.5).astype(np.uint8)\n",
        "#pred2 = (att_unet_model.predict(img_test_input)[0,:,:,1] > 0.5).astype(np.uint8)\n",
        "# Visualize the result \n",
        "plt.figure(figsize=(15, 10))\n",
        "plt.subplot(231)\n",
        "plt.title('Testing Image')\n",
        "plt.imshow(img_test, cmap='gray')\n",
        "plt.subplot(232)\n",
        "plt.title('Testing Label')\n",
        "plt.imshow(test_label[:,:,:], cmap='gray')\n",
        "plt.subplot(233)\n",
        "plt.title('Prediction on test image')\n",
        "plt.imshow(pred, cmap='gray')\n",
        "\n",
        "\n",
        "num_img_test+=1\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cw0ExfuOzxAx"
      },
      "source": [
        "# MODEL 7"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8IJHShKNz0B5",
        "outputId": "0223a161-6ecc-4b8e-8371-e007ce51b2a9"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Collecting keras-unet-collection\n",
            "  Downloading keras_unet_collection-0.1.11-py3-none-any.whl (67 kB)\n",
            "\u001b[?25l\r\u001b[K     |████▉                           | 10 kB 34.0 MB/s eta 0:00:01\r\u001b[K     |█████████▊                      | 20 kB 25.3 MB/s eta 0:00:01\r\u001b[K     |██████████████▌                 | 30 kB 29.3 MB/s eta 0:00:01\r\u001b[K     |███████████████████▍            | 40 kB 28.6 MB/s eta 0:00:01\r\u001b[K     |████████████████████████▏       | 51 kB 17.4 MB/s eta 0:00:01\r\u001b[K     |█████████████████████████████   | 61 kB 19.3 MB/s eta 0:00:01\r\u001b[K     |████████████████████████████████| 67 kB 4.8 MB/s \n",
            "\u001b[?25hInstalling collected packages: keras-unet-collection\n",
            "Successfully installed keras-unet-collection-0.1.11\n"
          ]
        }
      ],
      "source": [
        "!pip install keras-unet-collection"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "DR1SxJYSz0y2"
      },
      "outputs": [],
      "source": [
        "from keras_unet_collection import models,utils\n",
        "import numpy as np\n",
        "from glob import glob\n",
        "from tensorflow import keras \n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 232
        },
        "id": "AmIuh5KWz-2l",
        "outputId": "53c0d275-c8e8-4ed4-e721-7ea31884f838"
      },
      "outputs": [
        {
          "ename": "NameError",
          "evalue": "ignored",
          "output_type": "error",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-25-f996156aad39>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m model = swin_unet_2d((256, 256, 3), filter_num=[64, 128, 256, 512, 1024], n_labels=3, \n\u001b[0m\u001b[1;32m      2\u001b[0m                            \u001b[0mstack_num_down\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mstack_num_up\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mactivation\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'ReLU'\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m                            \u001b[0matten_activation\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'ReLU'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mattention\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'add'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moutput_activation\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'Softmax'\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m                            \u001b[0mpool\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0munpool\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m                            \u001b[0mbackbone\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'VGG16'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mweights\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'imagenet'\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'swin_unet_2d' is not defined"
          ]
        }
      ],
      "source": [
        "model = swin_unet_2d((256, 256, 3), filter_num=[64, 128, 256, 512, 1024], n_labels=3, \n",
        "                           stack_num_down=2, stack_num_up=2, activation='ReLU', \n",
        "                           atten_activation='ReLU', attention='add', output_activation='Softmax', \n",
        "                           pool=False, unpool=False, \n",
        "                           backbone='VGG16', weights='imagenet', \n",
        "                           freeze_backbone=True, freeze_batch_norm=True, \n",
        "                           name='attunet')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "H1h-q5nJ0tYD",
        "outputId": "91faa42e-733c-406c-f659-bb55a844abfb"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"attunet_model\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            [(None, 256, 256, 3) 0                                            \n",
            "__________________________________________________________________________________________________\n",
            "VGG16_backbone (Functional)     [(None, 256, 256, 64 14714688    input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_decode_trans_conv ( (None, 32, 32, 512)  2359808     VGG16_backbone[0][4]             \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_decode_activation ( (None, 32, 32, 512)  0           attunet_up0_decode_trans_conv[0][\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_att_theta_x (Conv2D (None, 32, 32, 256)  131328      VGG16_backbone[0][3]             \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_att_phi_g (Conv2D)  (None, 32, 32, 256)  131328      attunet_up0_decode_activation[0][\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_att_add (Add)       (None, 32, 32, 256)  0           attunet_up0_att_theta_x[0][0]    \n",
            "                                                                 attunet_up0_att_phi_g[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_att_activation (ReL (None, 32, 32, 256)  0           attunet_up0_att_add[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_att_psi_f (Conv2D)  (None, 32, 32, 1)    257         attunet_up0_att_activation[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_att_sigmoid (Activa (None, 32, 32, 1)    0           attunet_up0_att_psi_f[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_att_masking (Multip (None, 32, 32, 512)  0           VGG16_backbone[0][3]             \n",
            "                                                                 attunet_up0_att_sigmoid[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_concat (Concatenate (None, 32, 32, 1024) 0           attunet_up0_decode_activation[0][\n",
            "                                                                 attunet_up0_att_masking[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_conv_after_concat_0 (None, 32, 32, 512)  4719104     attunet_up0_concat[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_conv_after_concat_0 (None, 32, 32, 512)  0           attunet_up0_conv_after_concat_0[0\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_conv_after_concat_1 (None, 32, 32, 512)  2359808     attunet_up0_conv_after_concat_0_a\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up0_conv_after_concat_1 (None, 32, 32, 512)  0           attunet_up0_conv_after_concat_1[0\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_decode_trans_conv ( (None, 64, 64, 256)  1179904     attunet_up0_conv_after_concat_1_a\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_decode_activation ( (None, 64, 64, 256)  0           attunet_up1_decode_trans_conv[0][\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_att_theta_x (Conv2D (None, 64, 64, 128)  32896       VGG16_backbone[0][2]             \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_att_phi_g (Conv2D)  (None, 64, 64, 128)  32896       attunet_up1_decode_activation[0][\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_att_add (Add)       (None, 64, 64, 128)  0           attunet_up1_att_theta_x[0][0]    \n",
            "                                                                 attunet_up1_att_phi_g[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_att_activation (ReL (None, 64, 64, 128)  0           attunet_up1_att_add[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_att_psi_f (Conv2D)  (None, 64, 64, 1)    129         attunet_up1_att_activation[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_att_sigmoid (Activa (None, 64, 64, 1)    0           attunet_up1_att_psi_f[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_att_masking (Multip (None, 64, 64, 256)  0           VGG16_backbone[0][2]             \n",
            "                                                                 attunet_up1_att_sigmoid[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_concat (Concatenate (None, 64, 64, 512)  0           attunet_up1_decode_activation[0][\n",
            "                                                                 attunet_up1_att_masking[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_conv_after_concat_0 (None, 64, 64, 256)  1179904     attunet_up1_concat[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_conv_after_concat_0 (None, 64, 64, 256)  0           attunet_up1_conv_after_concat_0[0\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_conv_after_concat_1 (None, 64, 64, 256)  590080      attunet_up1_conv_after_concat_0_a\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up1_conv_after_concat_1 (None, 64, 64, 256)  0           attunet_up1_conv_after_concat_1[0\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_decode_trans_conv ( (None, 128, 128, 128 295040      attunet_up1_conv_after_concat_1_a\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_decode_activation ( (None, 128, 128, 128 0           attunet_up2_decode_trans_conv[0][\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_att_theta_x (Conv2D (None, 128, 128, 64) 8256        VGG16_backbone[0][1]             \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_att_phi_g (Conv2D)  (None, 128, 128, 64) 8256        attunet_up2_decode_activation[0][\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_att_add (Add)       (None, 128, 128, 64) 0           attunet_up2_att_theta_x[0][0]    \n",
            "                                                                 attunet_up2_att_phi_g[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_att_activation (ReL (None, 128, 128, 64) 0           attunet_up2_att_add[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_att_psi_f (Conv2D)  (None, 128, 128, 1)  65          attunet_up2_att_activation[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_att_sigmoid (Activa (None, 128, 128, 1)  0           attunet_up2_att_psi_f[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_att_masking (Multip (None, 128, 128, 128 0           VGG16_backbone[0][1]             \n",
            "                                                                 attunet_up2_att_sigmoid[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_concat (Concatenate (None, 128, 128, 256 0           attunet_up2_decode_activation[0][\n",
            "                                                                 attunet_up2_att_masking[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_conv_after_concat_0 (None, 128, 128, 128 295040      attunet_up2_concat[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_conv_after_concat_0 (None, 128, 128, 128 0           attunet_up2_conv_after_concat_0[0\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_conv_after_concat_1 (None, 128, 128, 128 147584      attunet_up2_conv_after_concat_0_a\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up2_conv_after_concat_1 (None, 128, 128, 128 0           attunet_up2_conv_after_concat_1[0\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_decode_trans_conv ( (None, 256, 256, 64) 73792       attunet_up2_conv_after_concat_1_a\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_decode_activation ( (None, 256, 256, 64) 0           attunet_up3_decode_trans_conv[0][\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_att_theta_x (Conv2D (None, 256, 256, 32) 2080        VGG16_backbone[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_att_phi_g (Conv2D)  (None, 256, 256, 32) 2080        attunet_up3_decode_activation[0][\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_att_add (Add)       (None, 256, 256, 32) 0           attunet_up3_att_theta_x[0][0]    \n",
            "                                                                 attunet_up3_att_phi_g[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_att_activation (ReL (None, 256, 256, 32) 0           attunet_up3_att_add[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_att_psi_f (Conv2D)  (None, 256, 256, 1)  33          attunet_up3_att_activation[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_att_sigmoid (Activa (None, 256, 256, 1)  0           attunet_up3_att_psi_f[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_att_masking (Multip (None, 256, 256, 64) 0           VGG16_backbone[0][0]             \n",
            "                                                                 attunet_up3_att_sigmoid[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_concat (Concatenate (None, 256, 256, 128 0           attunet_up3_decode_activation[0][\n",
            "                                                                 attunet_up3_att_masking[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_conv_after_concat_0 (None, 256, 256, 64) 73792       attunet_up3_concat[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_conv_after_concat_0 (None, 256, 256, 64) 0           attunet_up3_conv_after_concat_0[0\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_conv_after_concat_1 (None, 256, 256, 64) 36928       attunet_up3_conv_after_concat_0_a\n",
            "__________________________________________________________________________________________________\n",
            "attunet_up3_conv_after_concat_1 (None, 256, 256, 64) 0           attunet_up3_conv_after_concat_1[0\n",
            "__________________________________________________________________________________________________\n",
            "attunet_output (Conv2D)         (None, 256, 256, 3)  195         attunet_up3_conv_after_concat_1_a\n",
            "__________________________________________________________________________________________________\n",
            "attunet_output_activation (Soft (None, 256, 256, 3)  0           attunet_output[0][0]             \n",
            "==================================================================================================\n",
            "Total params: 28,375,271\n",
            "Trainable params: 13,660,583\n",
            "Non-trainable params: 14,714,688\n",
            "__________________________________________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "qQU7Rb0O26sd"
      },
      "outputs": [],
      "source": [
        "from keras_unet_collection.losses import iou_seg\n",
        "\n",
        "model.compile(loss='categorical_crossentropy', optimizer=keras.optimizers.Adam(),metrics=['accuracy',iou_seg])\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "19Igip8Qe5oA"
      },
      "outputs": [],
      "source": [
        "model.load_weights(\"/content/drive/MyDrive/sotaSegMod/unetCollection.h5\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "EpDDfBxj4Tef"
      },
      "outputs": [],
      "source": [
        "checkpoint_filepath=\"/content/drive/MyDrive/sotaSegMod/unetCollection2.h5\"\n",
        "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
        "    filepath=checkpoint_filepath,\n",
        "    save_weights_only=True,\n",
        "    monitor='val_iou_seg',\n",
        "    mode='min',\n",
        "    save_best_only=True)\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "XQGNm4Py2-dv"
      },
      "outputs": [],
      "source": [
        "history = model.fit(mySet.trainingArrayX, mySet.trainingArrayY, \n",
        "                             verbose=1,\n",
        "                             batch_size = 32,\n",
        "                             validation_data=(mySet.testArrayX,mySet.testArrayY),\n",
        "                             shuffle=False,\n",
        "                             epochs=50,\n",
        "                             callbacks=[model_checkpoint_callback])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "o5_SH0N7ffWU"
      },
      "outputs": [],
      "source": [
        "for i in range(len(mySet.nameArray)):\n",
        "  img_test = mySet.testArrayX[i]\n",
        "  img_test_input = np.expand_dims(img_test, 0)\n",
        "  plt.imsave(\"/content/testFolder/\"+str(mySet.nameArray[i]),(((model.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "J9gNfFMz91sX"
      },
      "outputs": [],
      "source": [
        "num_img_test=0"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 255
        },
        "id": "G1GVpK9g9ux3",
        "outputId": "ea913bc7-a89d-4089-9450-35d99f513b07"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1080x720 with 3 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "\n",
        "img_test = mySet.testArrayX[num_img_test]\n",
        "test_label = mySet.testArrayY[num_img_test]\n",
        "\n",
        "img_test_input = np.expand_dims(img_test, 0)\n",
        "\n",
        "pred = ((model.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "'''pred2 = ((att_unet_model2.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred3 = ((att_unet_model3.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred4 = ((att_unet_model4.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred5 = ((att_unet_model5.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)'''\n",
        "#pred = (att_unet_model.predict(img_test_input)[0,:,:,0] > 0.5).astype(np.uint8)\n",
        "#pred2 = (att_unet_model.predict(img_test_input)[0,:,:,1] > 0.5).astype(np.uint8)\n",
        "# Visualize the result \n",
        "plt.figure(figsize=(15, 10))\n",
        "plt.subplot(231)\n",
        "plt.title('Testing Image')\n",
        "plt.imshow(img_test, cmap='gray')\n",
        "plt.subplot(232)\n",
        "plt.title('Testing Label')\n",
        "plt.imshow(test_label[:,:,:], cmap='gray')\n",
        "plt.subplot(233)\n",
        "plt.title('Prediction on test image')\n",
        "plt.imshow(pred, cmap='gray')\n",
        "\n",
        "\n",
        "num_img_test+=1"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GPm3eOQT1FWP"
      },
      "source": [
        "# MODEL 8"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hinV8tHU1Gt0",
        "outputId": "c633c88b-a389-4b3e-a2dd-fde14aa9e222"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/keras_unet_collection/_backbone_zoo.py:45: UserWarning: \n",
            "\n",
            "Backbone VGG16 does not use batch norm, but other layers received batch_norm=True\n",
            "  warnings.warn(param_mismatch);\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg16/vgg16_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
            "58892288/58889256 [==============================] - 0s 0us/step\n",
            "58900480/58889256 [==============================] - 0s 0us/step\n"
          ]
        }
      ],
      "source": [
        "model_Unet = models.unet_2d((256, 256, 3), filter_num=[64, 128, 256, 512, 1024], \n",
        "                           n_labels=3, \n",
        "                           stack_num_down=2, stack_num_up=2, \n",
        "                           activation='ReLU', \n",
        "                           output_activation='Softmax', \n",
        "                           batch_norm=True, pool=False, unpool=False, \n",
        "                           backbone='VGG16', weights='imagenet', \n",
        "                           freeze_backbone=True, freeze_batch_norm=True, \n",
        "                           name='unet')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "2kKeRZMa1Lfe"
      },
      "outputs": [],
      "source": [
        "model_Unet.load_weights(\"/content/drive/MyDrive/sotaSegMod/model_Unet2D.h5\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "y1Cgf5F01oIT"
      },
      "outputs": [],
      "source": [
        "from keras_unet_collection.losses import iou_seg\n",
        "\n",
        "model_Unet.compile(loss='categorical_crossentropy', optimizer=keras.optimizers.Adam(learning_rate=0.0001),metrics=['accuracy',iou_seg])\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "82Aa14wb1v2-"
      },
      "outputs": [],
      "source": [
        "checkpoint_filepath=\"/content/drive/MyDrive/sotaSegMod/model_Unet2D.h5\"\n",
        "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
        "    filepath=checkpoint_filepath,\n",
        "    save_weights_only=True,\n",
        "    monitor='val_iou_seg',\n",
        "    mode='min',\n",
        "    save_best_only=True)\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YRcYPZgI1x3n",
        "outputId": "9a25144c-2786-4951-f77f-a6363570cbda"
      },
      "outputs": [],
      "source": [
        "history = model_Unet.fit(mySet.trainingArrayX, mySet.trainingArrayY, \n",
        "                             verbose=1,\n",
        "                             batch_size = 32,\n",
        "                             validation_data=(mySet.testArrayX,mySet.testArrayY),\n",
        "                             shuffle=False,\n",
        "                             epochs=200,\n",
        "                             callbacks=[model_checkpoint_callback])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "JPvOwGJB434N"
      },
      "outputs": [],
      "source": [
        "for i in range(len(mySet.nameArray)):\n",
        "  img_test = mySet.testArrayX[i]\n",
        "  img_test_input = np.expand_dims(img_test, 0)\n",
        "  plt.imsave(\"/content/testFolder/\"+str(mySet.nameArray[i]),(((model.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pZoxlg-9IxSb"
      },
      "source": [
        "# MODEL 9"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "B5oi5fNjIyx_",
        "outputId": "9be1979b-7926-4c80-ad6d-9600b447ad17"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/keras_unet_collection/_backbone_zoo.py:45: UserWarning: \n",
            "\n",
            "Backbone VGG16 does not use batch norm, but other layers received batch_norm=True\n",
            "  warnings.warn(param_mismatch);\n",
            "/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/optimizer_v2.py:356: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead.\n",
            "  \"The `lr` argument is deprecated, use `learning_rate` instead.\")\n"
          ]
        }
      ],
      "source": [
        "from keras_unet_collection import losses\n",
        "model = models.att_unet_2d((256, 256, 3), filter_num=[64, 128, 256, 512, 1024], n_labels=2, \n",
        "                           stack_num_down=4, stack_num_up=2, activation='ReLU', \n",
        "                           atten_activation='ReLU', attention='add', output_activation='Sigmoid', \n",
        "                           batch_norm=True, pool=True, unpool=False, \n",
        "                           backbone='VGG16', weights='imagenet', \n",
        "                           freeze_backbone=False, freeze_batch_norm=False, \n",
        "                           name='attunet')\n",
        "\n",
        "model.compile(loss='categorical_crossentropy', optimizer=tf.keras.optimizers.Adam(lr = 1e-3), \n",
        "              metrics=['accuracy', losses.dice_coef,losses.iou_seg])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "jeG115807uE4"
      },
      "outputs": [],
      "source": [
        "help(models.att_unet_2d)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "FzrisYnzJAWk"
      },
      "outputs": [],
      "source": [
        "checkpoint_filepath=\"/content/drive/MyDrive/collectionModels/attUnet.h5\"\n",
        "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
        "    filepath=checkpoint_filepath,\n",
        "    save_weights_only=True,\n",
        "    monitor='val_iou_seg',\n",
        "    mode='min',\n",
        "    save_best_only=True)\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "-WxYhmAjTPdy"
      },
      "outputs": [],
      "source": [
        "tf.keras.utils.plot_model(model)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 358
        },
        "id": "rOtLYjYMJVKn",
        "outputId": "70275cca-c432-434d-a9ef-7aa191a132f5"
      },
      "outputs": [],
      "source": [
        "history = model_att_unet.fit(mySet.trainingArrayX, mySet.trainingArrayY, \n",
        "                             verbose=1,\n",
        "                             batch_size = 32,\n",
        "                             validation_data=(mySet.testArrayX,mySet.testArrayY),\n",
        "                             shuffle=True,\n",
        "                             epochs=100,\n",
        "                             callbacks=[model_checkpoint_callback])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "HyNhlr-9NpxF"
      },
      "outputs": [],
      "source": [
        "for i in range(len(mySet.nameArray)):\n",
        "  img_test = mySet.testArrayX[i]\n",
        "  img_test_input = np.expand_dims(img_test, 0)\n",
        "  plt.imsave(\"/content/testFolder/\"+str(mySet.nameArray[i]),(((model_att_unet.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "B0x0rpa34JUC"
      },
      "source": [
        "# MODEL 10"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "5mNknq_V4Sqz"
      },
      "outputs": [],
      "source": [
        "from keras_unet_collection import models,utils\n",
        "import numpy as np\n",
        "from glob import glob\n",
        "from tensorflow import keras \n",
        "from keras_unet_collection.losses import iou_seg,dice_coef\n",
        "from keras import backend as K"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "HxJyrUnU4PTi"
      },
      "outputs": [],
      "source": [
        "model = models.unet_2d((256, 256, 3), [512,1024], n_labels=3,\n",
        "                      stack_num_down=2, stack_num_up=1,\n",
        "                      activation='GELU', output_activation='Softmax', \n",
        "                      batch_norm=True, pool='max', unpool='nearest', name='unet')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hLI62i2F_SqT",
        "outputId": "613f31f7-135d-4153-9f3a-4a1f5fd91963"
      },
      "outputs": [],
      "source": [
        "model.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "KwuQUYVo-lb_"
      },
      "outputs": [],
      "source": [
        "focal_loss = sm.losses.CategoricalCELoss()\n",
        "\n",
        "adam=tf.keras.optimizers.Adam(0.00005)\n",
        "metric = [sm.metrics.IOUScore(threshold=0.5), sm.metrics.FScore(threshold=0.5)]\n",
        "model.compile(loss=focal_loss, optimizer=adam, \n",
        "              metrics=metric)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 307
        },
        "id": "AVF_tOjszSoO",
        "outputId": "9746fc7c-5176-4ed9-de76-f89b62ec3ffa"
      },
      "outputs": [],
      "source": [
        "model.load_weights(\"/content/drive/MyDrive/collectionModels/augmentModeladam.h5\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "1C6K3nQy-vyQ"
      },
      "outputs": [],
      "source": [
        "checkpoint_filepath=\"/content/drive/MyDrive/collectionModels/enerjik.h5\"\n",
        "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
        "    filepath=checkpoint_filepath,\n",
        "    save_weights_only=True,\n",
        "    monitor='val_iou_score',\n",
        "    mode='max',\n",
        "    save_best_only=True)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 375
        },
        "id": "q-89kVJv-2fU",
        "outputId": "a162acf2-a641-46c7-f016-1f5ea0ac7bcb"
      },
      "outputs": [],
      "source": [
        "history = model.fit(mySet.trainingArrayX, mySet.trainingArrayY, \n",
        "                             verbose = 1,\n",
        "                             batch_size = 4,\n",
        "                             validation_data = (mySet.testArrayX,mySet.testArrayY),\n",
        "                             shuffle = False,\n",
        "                             epochs = 25,\n",
        "                             callbacks = [model_checkpoint_callback])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "dXR1nZ9QUlQp"
      },
      "outputs": [],
      "source": [
        "from matplotlib import pyplot as plt\n",
        "import cv2\n",
        "for i in range(len(mySet.nameArray)):\n",
        "  img_test = mySet.testArrayX[i]\n",
        "  img_test_input = np.expand_dims(img_test, 0)\n",
        "  plt.imsave(\"/content/testFolder/\"+str(mySet.nameArray[i]),cv2.cvtColor(((model.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8),cv2.COLOR_BGR2RGB))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "YU39ix8pxv0l"
      },
      "outputs": [],
      "source": [
        "modelTwo = models.unet_2d((256, 256, 3), [64, 128, 256, 512, 1024], n_labels=3,\n",
        "                      stack_num_down=2, stack_num_up=2,\n",
        "                      activation='ReLU', output_activation='Softmax', \n",
        "                      batch_norm=True, pool='max', unpool='nearest', name='unet')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "UFBvOWlMx1RE"
      },
      "outputs": [],
      "source": [
        "modelTwo.compile(loss='categorical_crossentropy', optimizer=tf.keras.optimizers.Adam(), \n",
        "              metrics=[dice_coef,iou_seg])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "1tokCnANx4Dm"
      },
      "outputs": [],
      "source": [
        "modelTwo.load_weights(\"/content/drive/MyDrive/collectionModels/unetReluOne.h5\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2Sx41KIy9YCb"
      },
      "source": [
        "# MODEL 11"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "cSIaKrSZ9W6A"
      },
      "outputs": [],
      "source": [
        "!pip install -U segmentation-models"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "id": "tQEBgjWhLeJG",
        "outputId": "419305d7-5a34-40e9-f0b0-3f7c2083643b"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Segmentation Models: using `keras` framework.\n"
          ]
        },
        {
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'tf.keras'"
            ]
          },
          "execution_count": 7,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "import segmentation_models as sm\n",
        "sm.set_framework('tf.keras')\n",
        "sm.framework()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "B5kCXSR09gMh",
        "outputId": "b4777547-33c5-4d27-fd12-70e31b1966de"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "2.6.0\n",
            "2.6.0\n"
          ]
        }
      ],
      "source": [
        "import keras\n",
        "import tensorflow as tf\n",
        "print(keras.__version__)\n",
        "print(tf.__version__)\n",
        "#from keras.utils import normalize\n",
        "import os\n",
        "import glob\n",
        "import cv2\n",
        "import numpy as np\n",
        "from matplotlib import pyplot as plt\n",
        "from segmentation_models.losses import cce_jaccard_loss\n",
        "from segmentation_models.metrics import iou_score"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "VHZqC5g496uf"
      },
      "outputs": [],
      "source": [
        "from tensorflow.keras.layers import Input,Conv2D,Rescaling\n",
        "from tensorflow.keras import Model\n",
        "base_model = sm.Unet('vgg16', classes=3, activation='softmax',encoder_weights='imagenet',decoder_use_batchnorm=True)\n",
        "inp = Input(shape=(None, None, 1))\n",
        "l1 = Conv2D(3, (1, 1))(inp) # map N channels data to 3 channels\n",
        "\n",
        "out = base_model(l1)\n",
        "\n",
        "\n",
        "model = Model(inp, out, name=base_model.name)\n",
        "#model=base_model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ACSDtD4SBJrB"
      },
      "outputs": [],
      "source": [
        "model.load_weights(\"/content/drive/MyDrive/SegmentModelLastDay/vgg256unetilkoturum.h5\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "JSZRno_bchpT",
        "outputId": "8e046ed5-15c0-4ba7-f7e2-543ae91f630d"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7efde0103690>"
            ]
          },
          "execution_count": 12,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.imshow(mySet.trainingArrayY[0][:,:,0])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "YrtlcswPd8DJ"
      },
      "outputs": [],
      "source": [
        "#model.load_weights('/content/drive/MyDrive/ModelsForCompetition/mehmetakifgece.h5')\n",
        "\n",
        "dice_loss = sm.losses.DiceLoss(class_indexes=np.array([0,1])) \n",
        "focal_loss = sm.losses.CategoricalCELoss(class_indexes=np.array([0,1]))\n",
        "jaccard_loss=sm.losses.JaccardLoss(class_indexes=np.array([0,1]))\n",
        "total_loss = 0.5*dice_loss + 0.25*jaccard_loss + 0.25*focal_loss\n",
        "adam=tf.keras.optimizers.Adam(learning_rate=0.0001) \n",
        "metrics = [sm.metrics.IOUScore(threshold=0.5,class_indexes=np.array([0,1])),sm.metrics.FScore(threshold=0.5,class_indexes=np.array([0,1]))]\n",
        "model.compile(optimizer=adam,loss=dice_loss,metrics=metrics)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "SOpkY7v_sZFs"
      },
      "outputs": [],
      "source": [
        "model.load_weights('/content/drive/MyDrive/SegmentModelLastDay/vgg256unetilkoturum.h5')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EYucjnCznkxa",
        "outputId": "c7298d94-17d9-4929-cd02-610d415eb066"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/keras/utils/generic_utils.py:497: CustomMaskWarning: Custom mask layers require a config and must override get_config. When loading, the custom mask layer must be passed to the custom_objects argument.\n",
            "  category=CustomMaskWarning)\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Saved model to disk\n"
          ]
        }
      ],
      "source": [
        "model_json = model.to_json()\n",
        "with open(\"/content/drive/MyDrive/SegmentationLastModels/modelResNetFocalTwo.json\", \"w\") as json_file:\n",
        "    json_file.write(model_json)\n",
        "# serialize weights to HDF5\n",
        "model.save_weights(\"/content/drive/MyDrive/SegmentationLastModels/modelResNetFocalTwo.h5\")\n",
        "print(\"Saved model to disk\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "-pOrsZ0Ten_n"
      },
      "outputs": [],
      "source": [
        "checkpoint_filepath = '/content/drive/MyDrive/SegmentModelLastDay/vgg256focalilkoturumiki.h5'\n",
        "model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
        "    filepath=checkpoint_filepath,\n",
        "    monitor='val_iou_score',\n",
        "    mode='max',\n",
        "    save_best_only=True,\n",
        "    save_weights_only=True)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "gr2mRwJNersn"
      },
      "outputs": [],
      "source": [
        "model.fit(\n",
        "   x=mySet.trainingArrayX,\n",
        "   y=mySet.trainingArrayY,\n",
        "   batch_size=2,\n",
        "   epochs=200,\n",
        "   validation_data=(mySet.testArrayX, mySet.testArrayY),\n",
        "   callbacks=[model_checkpoint_callback],\n",
        ")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "QHdwE64XYgCS"
      },
      "outputs": [],
      "source": [
        "vmodel.fit(\n",
        "   x=mySet.trainingArrayX,\n",
        "   y=mySet.trainingArrayY,\n",
        "   batch_size=4,\n",
        "   epochs=100,\n",
        "   validation_data=(mySet.testArrayX, mySet.testArrayY),\n",
        "   callbacks=[model_checkpoint_callback],\n",
        ")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "KoXuKPpeetjX"
      },
      "outputs": [],
      "source": [
        "for i in range(len(mySet.nameArray)):\n",
        "  img_test = mySet.testArrayX[i]\n",
        "  img_test_input = np.expand_dims(img_test, 0)\n",
        "  plt.imsave(\"/content/testFolder/\"+str(mySet.nameArray[i]),((model.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "QCUsrWdwlvpe"
      },
      "outputs": [],
      "source": [
        "model.save_weights(\"/content/drive/MyDrive/SegmentModelLastDay/79verenmodelsonepoch.h5\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "D0Recq7dDVPG"
      },
      "source": [
        "# Evaluation"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "o9rU3-32v_2-",
        "outputId": "4d579946-c856-4fd8-91a6-8a1b4c9daae0"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Loaded model from disk\n"
          ]
        }
      ],
      "source": [
        "from keras.models import model_from_json\n",
        "json_file = open('/content/drive/MyDrive/SegmentationLastModels/modelResNetFocalTwo.json', 'r')\n",
        "loaded_model_json = json_file.read()\n",
        "json_file.close()\n",
        "loaded_model = model_from_json(loaded_model_json)\n",
        "# load weights into new model\n",
        "loaded_model.load_weights(\"/content/drive/MyDrive/SegmentationLastModels/modelResNetFocalTwo.h5\")\n",
        "print(\"Loaded model from disk\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "8cRlfsWVwJdN"
      },
      "outputs": [],
      "source": [
        "#model.load_weights('/content/drive/MyDrive/ModelsForCompetition/mehmetakifgece.h5')\n",
        "dice_loss = sm.losses.DiceLoss(class_weights=np.array([1, 1, 0.5])) \n",
        "focal_loss = sm.losses.CategoricalFocalLoss()\n",
        "total_loss = (0.75 * dice_loss) + (0.25 * focal_loss)\n",
        "adam=tf.keras.optimizers.Adam(0.0001)\n",
        "metrics = [sm.metrics.IOUScore(threshold=0.5),sm.metrics.FScore(class_indexes=np.array([0,1]),threshold=0.5)]\n",
        "loaded_model.compile(optimizer=adam,loss=total_loss,metrics=metrics)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8tn2PoHFwq14",
        "outputId": "86f11404-1b77-4468-c7aa-32f1c7e661b7"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "14/14 [==============================] - 11s 391ms/step - loss: 0.2218 - iou_score: 0.7531 - f1-score: 0.8358\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "[0.22175036370754242, 0.7531320452690125, 0.8358188271522522]"
            ]
          },
          "execution_count": 12,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "model.evaluate(mySet.testArrayX,mySet.testArrayY)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "tJ4w1YcVDiZa"
      },
      "outputs": [],
      "source": [
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "coefList=[]\n",
        "lenlist=[]\n",
        "reals,preds,preds2=[],[],[]\n",
        "for i in range(len(mySet.nameArray)):\n",
        "   \n",
        "  predimg=cv2.imread(\"/content/drive/MyDrive/Segment/BoneNewTwo/TestSet/Annotation/\"+mySet.nameArray[i])\n",
        "  reals.append(predimg)\n",
        "  predimgOne=predimg[:,:,0:2]\n",
        "  realimg=cv2.resize(cv2.imread(\"/content/testFolder/\"+mySet.nameArray[i]),(512,512),interpolation=cv2.INTER_CUBIC)\n",
        "  realimg2=cv2.resize(cv2.imread(\"/content/testFolder/\"+mySet.nameArray[i]),(512,512),interpolation=cv2.INTER_NEAREST)\n",
        "  preds.append(realimg)\n",
        "  preds2.append(realimg2)\n",
        "  realimgOne=realimg2[:,:,0:2]\n",
        "  predimgOneF=predimgOne.flatten()\n",
        "  realimgOneF=realimgOne.flatten()\n",
        "  intersectOne=predimgOneF*realimgOneF\n",
        "  intersectOne=np.sum(intersectOne)\n",
        "\n",
        "  xorOne=np.sum(np.logical_xor(predimgOneF,realimgOneF))\n",
        "  iouOne=intersectOne/(intersectOne+xorOne)\n",
        "  coefList.append(iouOne)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GKJywUtok7kr",
        "outputId": "e87435c8-3f2e-41f1-e486-147a3ab92f9e"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "0.5241832048825514"
            ]
          },
          "execution_count": 13,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "sum(coefList)/len(coefList)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 232
        },
        "id": "e3h8KwWOt4DX",
        "outputId": "b028290b-15d4-46be-dfff-27bd7522dbfc"
      },
      "outputs": [],
      "source": [
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "coefList=[]\n",
        "lenlist=[]\n",
        "reals,preds,preds2=[],[],[]\n",
        "for i in range(len(mySet.nameArray)):\n",
        "   \n",
        "  predimg=cv2.imread(\"/content/drive/MyDrive/Segment/BoneNewTwo/TestSet/Annotation/\"+mySet.nameArray[i])\n",
        "  reals.append(predimg)\n",
        "  predimgOne=predimg[:,:,0:2]\n",
        "  realimg=cv2.cvtColor(cv2.imread(\"/content/testFolder/\"+mySet.nameArray[i]),cv2.COLOR_BGR2RGB)\n",
        "  realimg2=cv2.imread(\"/content/testFolder/\"+mySet.nameArray[i])\n",
        "  preds.append(realimg)\n",
        "  preds2.append(realimg2)\n",
        "  realimgOne=realimg[:,:,0:2]\n",
        "  predimgOneF=predimgOne.flatten()\n",
        "  realimgOneF=realimgOne.flatten()\n",
        "  intersectOne=predimgOneF*realimgOneF\n",
        "  intersectOne=np.sum(intersectOne)\n",
        "\n",
        "  xorOne=np.sum(np.logical_xor(predimgOneF,realimgOneF))\n",
        "  iouOne=intersectOne/(intersectOne+xorOne)\n",
        "  coefList.append(iouOne)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OtvCGxiuEjXv",
        "outputId": "48be18d3-4cdd-4373-8838-56d1f76aac0d"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "0.5855659877878219"
            ]
          },
          "execution_count": 19,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "sum(coefList)/len(coefList)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ghAovmjeJ7AD",
        "outputId": "760375ca-3f2d-43c1-8658-970effcec870"
      },
      "outputs": [],
      "source": [
        "import time\n",
        "scores=[]\n",
        "\n",
        "for i in range(len(mySet.nameArray)):\n",
        "    start = time.time()\n",
        "   \n",
        "    \n",
        "    grd=cv2.cvtColor(cv2.imread(\"/content/drive/MyDrive/Segment/BoneNewTwo/TestSet/Annotation/\"+mySet.nameArray[i]),cv2.COLOR_BGR2RGB)\n",
        "    prd=cv2.cvtColor(cv2.resize(cv2.imread(\"/content/testFolder/\"+mySet.nameArray[i]),(512,512),interpolation=cv2.INTER_CUBIC),cv2.COLOR_BGR2RGB)\n",
        "    newimage=np.zeros((512,512))\n",
        "    newimage2=np.zeros((512,512))\n",
        "    for i in range(512):\n",
        "      for j in range(512):\n",
        "        class_labels = np.argmax(prd[i,j], axis=0)\n",
        "        newimage[i][j]=class_labels\n",
        "        class_labelsTwo = np.argmax(grd[i,j], axis=0)\n",
        "        newimage2[i][j]=class_labelsTwo\n",
        "    predictedMask=newimage\n",
        "    groundtruthMask=newimage2\n",
        "    \n",
        "    iou=calculateIoU(groundtruthMask,predictedMask,False)\n",
        "    end = time.time()\n",
        "  \n",
        "    scores.append(iou)\n",
        "    print(end - start,iou,sum(scores)/len(scores))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OK88mmovZQLL",
        "outputId": "8b274d56-fe1b-4062-92da-33a1f084c69e"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "0.695549726698729"
            ]
          },
          "execution_count": 21,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "sum(scores)/len(scores)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 232
        },
        "id": "QV4pwLSY7ZPn",
        "outputId": "7fee2e9a-91a2-42bc-e0f4-68ca80d72f42"
      },
      "outputs": [],
      "source": [
        "scores=[]\n",
        "for i in range(len(mySet.nameArray)):\n",
        "\n",
        "    prd=cv2.imread(\"/content/drive/MyDrive/Segment/BoneNewTwo/TestSet/Annotation/\"+mySet.nameArray[i])\n",
        "    grd=cv2.resize(cv2.imread(\"/content/testFolder/\"+mySet.nameArray[i]),(512,512),interpolation=cv2.INTER_CUBIC)\n",
        "    newimage=np.zeros((512,512))\n",
        "    newimage2=np.zeros((512,512))\n",
        "    for i in range(512):\n",
        "      for j in range(512):\n",
        "        if prd[i][j][0]==255:\n",
        "          newimage[i][j]=1\n",
        "        if prd[i][j][1]==255:\n",
        "          newimage[i][j]=2\n",
        "        if prd[i][j][2]==255:\n",
        "          newimage[i][j]=0\n",
        "        if grd[i][j][0]==255:\n",
        "          newimage2[i][j]=1\n",
        "        if grd[i][j][1]==255:\n",
        "          newimage2[i][j]=2\n",
        "        if grd[i][j][2]==255:\n",
        "          newimage2[i][j]=0\n",
        "    predictedMask=newimage\n",
        "    groundtruthMask=newimage2\n",
        "    \n",
        "    iou=calculateIoU(groundtruthMask,predictedMask,False)\n",
        "    print(iou)\n",
        "    scores.append(iou)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JXplxLrABRqk",
        "outputId": "a1a5beda-f996-4ef2-d175-f6573991fcaf"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "0.7188215406567501"
            ]
          },
          "execution_count": 22,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "sum(scores)/len(scores)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "lvyXNBilNxCg"
      },
      "outputs": [],
      "source": [
        "\n",
        "newimage=np.zeros((512,512))\n",
        "for i in range(len(prd[0])):\n",
        "  print(prd[0][i])\n",
        "  if prd[0][i][0]==255:\n",
        "    newimage[i][i]=0\n",
        "  if prd[0][i][1]==255:\n",
        "    newimage[i][j]=1\n",
        "  if prd[0][i][2]==255:\n",
        "    newimage[i][j]=2\n",
        "  \n",
        "plt.imshow(newimage)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "9VI2ard1ul4v"
      },
      "outputs": [],
      "source": [
        "def loadMask(path):\n",
        "    reader = sitk.ImageFileReader()\n",
        "    reader.SetFileName(path)\n",
        "    sitkMask = reader.Execute();\n",
        "    mask = sitk.GetArrayFromImage(sitkMask)\n",
        "    \n",
        "    return mask\n",
        "\n",
        "def showGrayScale(mask, title):\n",
        "    plt.imshow(mask, cmap=plt.cm.gray)\n",
        "    plt.title(title)\n",
        "    plt.show()\n",
        "\n",
        "# 8 bit ve tek kanallı, normal (0), iskemik (1) ve kanama (2) sınıflarından oluşan\n",
        "# groundTruthMask ve predictedMask'i parametre olarak alip predictedMask'i puanlayan fonksiyon.\n",
        "# Maskelerin her adımdakı değişiminlerini görmek için showSteps = True.\n",
        "def calculateIoU(groundTruthMask, predictedMask, showSteps = False):\n",
        "    \n",
        "    if showSteps:\n",
        "        showGrayScale(groundtruthMask, \"Ground Truth Maskesi\")\n",
        "        showGrayScale(predictedMask, \"Predicted Maskesi\")\n",
        "    \n",
        "    # 1 ve 2 maskelerini ayrı ayrı oluşturuyoruz (0 ve 255 değerlerinin olduğu birer maske olarak)\n",
        "    mask1 = np.where(groundtruthMask == 1, 1, 0).astype(np.uint8)\n",
        "    mask2 = np.where(groundtruthMask == 2, 1, 0).astype(np.uint8)\n",
        "    \n",
        "\n",
        "    if showSteps:\n",
        "        showGrayScale(mask1, \"Ground Truth Sınıf 1\")\n",
        "        showGrayScale(mask2, \"Ground Truth Sınıf 2\")\n",
        "        \n",
        "\n",
        "    # İki sınıfı ayrı ayrı dilate ve erode ediyoruz\n",
        "    kernel = np.ones((3,3))\n",
        "\n",
        "    erosion1 = cv2.erode(mask1, kernel, iterations=1) \n",
        "    dilation1 = cv2.dilate(mask1, kernel, iterations=1)\n",
        "\n",
        "    erosion2 = cv2.erode(mask2, kernel, iterations=1) \n",
        "    dilation2 = cv2.dilate(mask2, kernel, iterations=1)\n",
        "    \n",
        "   \n",
        "    if showSteps:\n",
        "        showGrayScale(erosion1, \"Erode Edilmiş Ground Truth Sınıf 1\")\n",
        "    \n",
        "    \n",
        "    # Erode edilmiş ground truth class maskelerini birleştiriyoruz\n",
        "    erodedGroundtruth = np.zeros(groundtruthMask.shape, dtype = np.uint8)\n",
        "    erodedGroundtruth[erosion1 == 1] = 1\n",
        "    erodedGroundtruth[erosion2 == 1] = 2\n",
        "    \n",
        "    \n",
        "    if showSteps:\n",
        "        showGrayScale(erodedGroundtruth, \"Erode Edilmiş Ground Truthların Birleştirilmesi\")\n",
        "    \n",
        "    \n",
        "    # Dilate edilmiş ground truth class maskelerini birleştiriyoruz\n",
        "    dilatedGroundtruth = np.zeros(groundtruthMask.shape, dtype = np.uint8)\n",
        "    dilatedGroundtruth[dilation1 == 1] = 1\n",
        "    dilatedGroundtruth[dilation2 == 1] = 2    \n",
        "    \n",
        "    \n",
        "    if showSteps:\n",
        "        showGrayScale(dilatedGroundtruth, \"Dilate Edilmiş Ground Truthların Birleştirilmesi\")\n",
        "    \n",
        "    \n",
        "    # Dilate edilmiş ground truth ile kesişim\n",
        "    intersection = np.where(np.logical_and(dilatedGroundtruth == predictedMask, dilatedGroundtruth != 0), 1, 0)        \n",
        "    intersectionCount = np.count_nonzero(intersection)\n",
        "\n",
        "    # Erode edilmiş ground truth ile birleşim\n",
        "    union = np.where(np.logical_or(erodedGroundtruth != 0, predictedMask != 0), 1, 0)\n",
        "    unionCount = np.count_nonzero(union)\n",
        "\n",
        "    score = intersectionCount / unionCount\n",
        "\n",
        "    \n",
        "    if showSteps:\n",
        "        showGrayScale(intersection, \"Kesişim\")\n",
        "        showGrayScale(union, \"Birleşim\")\n",
        "        print('Kesişim piksel sayısı: ', intersectionCount)\n",
        "        print('Birleşim piksel sayısı: ', unionCount)\n",
        "        print('Puan: ', score)\n",
        "    \n",
        "    \n",
        "    return score"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "dLtYk_CmLKCJ"
      },
      "outputs": [],
      "source": [
        "num_img_test=27"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 310
        },
        "id": "rpDe-X5lLBhM",
        "outputId": "9cb430a2-9773-4f5f-c8d1-237c60d9cd1a"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1080x720 with 3 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "\n",
        "img_test = reals[num_img_test]\n",
        "test_label = preds2[num_img_test]\n",
        "pred = preds[num_img_test]\n",
        "'''pred2 = ((att_unet_model2.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred3 = ((att_unet_model3.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred4 = ((att_unet_model4.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)\n",
        "pred5 = ((att_unet_model5.predict(img_test_input)[0,:,:,:] > 0.5)*255).astype(np.uint8)'''\n",
        "#pred = (att_unet_model.predict(img_test_input)[0,:,:,0] > 0.5).astype(np.uint8)\n",
        "#pred2 = (att_unet_model.predict(img_test_input)[0,:,:,1] > 0.5).astype(np.uint8)\n",
        "# Visualize the result \n",
        "plt.figure(figsize=(15, 10))\n",
        "plt.subplot(231)\n",
        "\n",
        "plt.title('Testing Image')\n",
        "plt.imshow(img_test, cmap='gray')\n",
        "plt.subplot(232)\n",
        "plt.title('Testing Label')\n",
        "plt.imshow(test_label[:,:,:], cmap='gray')\n",
        "plt.subplot(233)\n",
        "plt.title('Prediction on test image')\n",
        "plt.imshow(pred, cmap='gray')\n",
        "\n",
        "num_img_test+=1"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "EJqy7Fv9EUiA",
        "outputId": "b72bc67d-3b18-4b2e-fc97-4cb1c1c605b8"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fea3f05b490>"
            ]
          },
          "execution_count": 75,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.imshow(test_label[:,:,2])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tUsWBMWGxUUK",
        "outputId": "8b88d864-9b02-4a56-e437-9906e407306d"
      },
      "outputs": [],
      "source": [
        "cv2.resize(imgm,(512,512),interpolation=cv2.INTER_NEAREST)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sGSLl8Nj6JU5"
      },
      "source": [
        "# Upload"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iI61wWYL681R",
        "outputId": "1eddbb26-c08b-44b6-e0c5-e254294fbc2c"
      },
      "outputs": [],
      "source": [
        "!pip install pydicom"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_8wQiEzlxVBA",
        "outputId": "31a4e3fb-3951-4422-d7ca-2ea31dec1124"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "4427"
            ]
          },
          "execution_count": 98,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "import pngGeneratorTwo\n",
        "gen=pngGeneratorTwo.PngGenerator(\"/content/drive/MyDrive/TRAINING/INMEYOK/DICOM\")\n",
        "gen.insertDicomData()\n",
        "len(gen.pathList)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rU9XRq9V694f",
        "outputId": "38bc9431-bbc1-4839-aa9c-2751838548fb"
      },
      "outputs": [],
      "source": [
        "gen.getPngData(\"/content/drive/MyDrive/CubicData\",\"Inmeyok\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lsZBl9XHTa1i"
      },
      "source": [
        "# WORKING CODE"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3mCusdYT7jvV",
        "outputId": "e883b575-33a8-4c54-c161-8f5cba6e3022"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: pydicom in /usr/local/lib/python3.7/dist-packages (2.2.1)\n"
          ]
        }
      ],
      "source": [
        "!pip install pydicom"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "8_69rIB3Mxpt"
      },
      "outputs": [],
      "source": [
        "from PIL import Image\n",
        "import numpy as np\n",
        "from numpy import asarray\n",
        "import cv2\n",
        "import tensorflow as tf\n",
        "import os\n",
        "from tensorflow.keras.models import model_from_json\n",
        "from tensorflow.keras.utils import to_categorical\n",
        "import pandas as pd"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "51Nj-HsbM1C3",
        "outputId": "0551cf9f-f117-44e3-edd1-82f5681c73a2"
      },
      "outputs": [],
      "source": [
        "import yarismaPngGenerator\n",
        "pngGen=yarismaPngGenerator.PngGenerator(\"/content/drive/MyDrive/yarismaOturumBir\",\"Classification\")\n",
        "pngGen.insertDicomData()\n",
        "pngGen.getPngData(\"/content/drive/MyDrive/pngOturumBir\",\"Data\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hV-snRfZNuGL",
        "outputId": "5573e34b-47be-45d8-af95-2f7375f3a34c"
      },
      "outputs": [],
      "source": [
        "# Import necessary components to build LeNet\n",
        "\n",
        "from keras.applications.vgg16 import VGG16\n",
        "from keras.layers import Dense, Dropout\n",
        "import numpy as np\n",
        "from keras.models import Model\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.models import Sequential\n",
        "from keras.applications.vgg16 import VGG16\n",
        "from keras.applications.xception import Xception\n",
        "from keras.applications.mobilenet import MobileNet\n",
        "from keras.applications.efficientnet import EfficientNetB5\n",
        "from keras.applications.inception_resnet_v2 import InceptionResNetV2\n",
        "from keras.applications.vgg19 import VGG19\n",
        "from keras.layers.core import Dense, Dropout, Activation, Flatten\n",
        "from keras.layers.convolutional import Conv2D, MaxPooling2D, ZeroPadding2D\n",
        "from keras.regularizers import l2\n",
        "import numpy as np\n",
        "from keras import layers\n",
        "from keras.layers import Input, Dense,Dropout, Activation,BatchNormalization, Flatten, Conv2D, MaxPooling2D\n",
        "from keras.models import Model\n",
        "from keras.preprocessing import image\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "import keras.backend as K\n",
        "import matplotlib.pyplot as plt\n",
        "from matplotlib.pyplot import imshow\n",
        "from keras.callbacks import EarlyStopping\n",
        "from keras.layers import Dense, GlobalAveragePooling2D\n",
        "\n",
        "from keras.layers import Dense, GlobalAveragePooling2D\n",
        "\n",
        "def vgg16():\n",
        "    \n",
        "  conv=VGG16(input_shape=[224,224,3],weights='imagenet',include_top=False)\n",
        "  conv.output\n",
        "\n",
        "\n",
        "  x = conv.output\n",
        "  x = Dropout(0.2)(x)\n",
        "  x = GlobalAveragePooling2D()(x)\n",
        "\n",
        "\n",
        "  x = Dense(2048,activation='relu')(x)\n",
        "  x = Dense(1024,activation='relu')(x)\n",
        "  x = Dropout(0.2)(x)\n",
        "  x = Dense(1024,activation='relu')(x)\n",
        "  x = Dense(512,activation='relu')(x)\n",
        "  \n",
        "  \n",
        "  pred = Dense(2,activation='softmax')(x)\n",
        "  model = Model(inputs = conv.input,outputs=pred)\n",
        "\n",
        "  model.summary()\n",
        "\n",
        "\n",
        "  for layer in conv.layers:\n",
        "      layer.trainable = False\n",
        "  return model\n",
        "model=vgg16()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "aYu4ZjCDNyB_"
      },
      "outputs": [],
      "source": [
        "adam=tf.keras.optimizers.Adam(learning_rate=0.00001)\n",
        "model.compile(optimizer = adam ,loss=\"categorical_crossentropy\", metrics=['accuracy'])\n",
        "model.load_weights(\"/content/drive/MyDrive/SagliktaYapayZekaModelleri/modelAugmented.h5\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "l3Kd5YFns3FS",
        "outputId": "cd0194dc-c4a3-4d38-ae92-217af6574704"
      },
      "outputs": [],
      "source": [
        "filenamelist"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "cNoq3xpjM4Tu"
      },
      "outputs": [],
      "source": [
        "liste=list()\n",
        "liste2=list()\n",
        "filenamelist=list()\n",
        "path=\"/content/drive/MyDrive/pngOturumBir\"\n",
        "predictions=[]\n",
        "for subdir,dirs,files in os.walk(path):\n",
        "  for file in files:\n",
        "    filePath=subdir+os.sep+file\n",
        "    img=cv2.imread(filePath)\n",
        "    img_test = cv2.resize(img,(224,224),interpolation=cv2.INTER_CUBIC)\n",
        "    liste.append(img_test)\n",
        "    liste2.append(0)\n",
        "    img_test_input = np.expand_dims(img_test,0)\n",
        "    img_test_input=img_test_input.astype(np.float32)\n",
        "    img_test_input=img_test_input/255.\n",
        "    strfile=file\n",
        "    filenamelist.append(strfile.split(\".\")[0])\n",
        "\n",
        "    prediction = model.predict(img_test_input)\n",
        "    prediction=(prediction>0.5).astype(np.uint8)\n",
        "    predictions.append(prediction)\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "i_XnIvyRN7Z_"
      },
      "outputs": [],
      "source": [
        "preds=list()\n",
        "\n",
        "for i in range(len(predictions)):\n",
        "  preds.append(int(predictions[i][0][1]))\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "id": "hfWecAhAN8xf",
        "outputId": "0f30c39b-1fc6-4e8e-90d7-9cc86ba7d5dc"
      },
      "outputs": [],
      "source": [
        "import pandas as pd\n",
        "\n",
        "data={\n",
        "    \"ID\": filenamelist,\n",
        "    \"ETİKET\": preds\n",
        "}\n",
        "\n",
        "df=pd.DataFrame(data)\n",
        "df = df.reset_index(drop=True)\n",
        "\n",
        "csv=df.to_csv(\"/content/drive/MyDrive/OTURUM1.csv\",index=False)\n",
        "\n",
        "df"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "WaTCXoGZ42Vc"
      },
      "outputs": [],
      "source": [
        "listeyeni=list()\n",
        "for i in pngGen.pathList:\n",
        " spltd=i.split(os.sep)[-1]\n",
        " spltdtwo=spltd.split(\".\")[0]\n",
        " if spltdtwo not in filenamelist:\n",
        "   listeyeni.append(spltdtwo)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AoFY1NvQ5Q0O",
        "outputId": "384045a2-0dd7-44f0-fefc-6198f88dd273"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "['11074']"
            ]
          },
          "execution_count": 15,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "listeyeni"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "vTVuay3l5R0Q"
      },
      "outputs": [],
      "source": []
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [
        "9BvpgZqIpT-W",
        "LQ_hz2_y8ORn",
        "B3aU82P0o2Pd",
        "GPm3eOQT1FWP"
      ],
      "machine_shape": "hm",
      "name": "Models For Segmentation.ipynb",
      "provenance": [],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}