5_training_mobileNetV3Large.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "gpuClass": "standard",
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "AIM: To train MobileNetV3Large model on different parameters."
      ],
      "metadata": {
        "id": "CnJiZ1Q-K0qs"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "TB7SxHDsUD2w"
      },
      "outputs": [],
      "source": [
        "# Note: Comment or delete the current code block if dataset is loaded from local disk.\n",
        "# loading dataset from kaggle\n",
        "# upload kaggle.json file\n",
        "! pip install -q kaggle\n",
        "from google.colab import files\n",
        "print(\"Upload 'kaggle.json' file...\")\n",
        "files.upload()\n",
        "\n",
        "# download the dataset(zipped folder) and unzip it\n",
        "! mkdir ~/.kaggle\n",
        "! cp kaggle.json ~/.kaggle/\n",
        "! chmod 600 ~/.kaggle/kaggle.json\n",
        "! kaggle datasets download alessiocorrado99/animals10\n",
        "! unzip animals10.zip"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Note: Load the class files into current working directory.\n",
        "# importing required libraries\n",
        "from tensorflow.keras.utils import image_dataset_from_directory\n",
        "import tensorflow as tf\n",
        "\n",
        "from keras.callbacks import ModelCheckpoint\n",
        "\n",
        "from imutils import paths\n",
        "import matplotlib.pyplot as plt\n",
        "import os\n",
        "\n",
        "import mobilenetv3l # class file\n",
        "import helper_functions # class file"
      ],
      "metadata": {
        "id": "OTJGpCW0UKgb"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# defining image size and batch size\n",
        "batch_size=64\n",
        "image_width=224\n",
        "image_height=224"
      ],
      "metadata": {
        "id": "jXiUug4QUvxi"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# getting output class labels and calculating class weight\n",
        "class_labels = sorted(os.listdir('raw-img')) # get class labels\n",
        "print('\\nClass Labels: {}'.format(class_labels))\n",
        "num_classes = len(class_labels) # get number of output classes\n",
        "print('\\nNumber of Classes: {}'.format(num_classes))\n",
        "\n",
        "# calculating total elements in each class\n",
        "class_totals = [len(os.listdir('raw-img' + f\"/{label}\")) for label in class_labels]\n",
        "print(\"\\nTotal Elements in Each Class : {}\".format(class_totals))\n",
        "\n",
        "# calculate class weight\n",
        "class_weight = {i:round((max(class_totals) / class_total), 3) for i, class_total in enumerate(class_totals)}\n",
        "print('\\nClass Weight: {}'.format(class_weight))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hOXNndS2LPoB",
        "outputId": "f27f8397-7180-4acd-ffaf-de9854970d7e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "Class Labels: ['cane', 'cavallo', 'elefante', 'farfalla', 'gallina', 'gatto', 'mucca', 'pecora', 'ragno', 'scoiattolo']\n",
            "\n",
            "Number of Classes: 10\n",
            "\n",
            "Total Elements in Each Class : [4863, 2623, 1446, 2112, 3098, 1668, 1866, 1820, 4821, 1862]\n",
            "\n",
            "Class Weight: {0: 1.0, 1: 1.854, 2: 3.363, 3: 2.303, 4: 1.57, 5: 2.915, 6: 2.606, 7: 2.672, 8: 1.009, 9: 2.612}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# get image paths and move them into train and validation directories\n",
        "imagePaths = list(paths.list_images(\"raw-img\")) # getting path to the directory of images\n",
        "print('Processing...')\n",
        "helper_functions.move_images(imagePaths=imagePaths) # calling function\n",
        "print('Completed...')"
      ],
      "metadata": {
        "id": "WRecGxWFx46_",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "27078e6a-182c-48c2-d078-56c5c2ebcc03"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Processing...\n",
            "Completed...\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# get training dataset\n",
        "train_ds = image_dataset_from_directory(\n",
        "  \"train_dir/\", label_mode='categorical', seed=42, shuffle=True,\n",
        "  image_size=(image_height, image_width),\n",
        "  batch_size=batch_size)\n",
        "\n",
        "# get validation dataset\n",
        "val_ds = image_dataset_from_directory(\n",
        "  \"valid_dir/\", label_mode='categorical', seed=42, shuffle=True,\n",
        "  image_size=(image_height, image_width),\n",
        "  batch_size=batch_size)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5mnkH8bNUyNp",
        "outputId": "a3182326-4cef-40d2-d967-f4de55b496d1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found 18325 files belonging to 10 classes.\n",
            "Found 7854 files belonging to 10 classes.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# using buffered prefetching\n",
        "shuffle_size = len(train_ds) # shuffles entire train dataset at each epoch\n",
        "\n",
        "AUTOTUNE = tf.data.AUTOTUNE\n",
        "train_ds = train_ds.cache().shuffle(shuffle_size).prefetch(buffer_size=AUTOTUNE)\n",
        "val_ds = val_ds.cache().prefetch(buffer_size=AUTOTUNE)"
      ],
      "metadata": {
        "id": "mpOT6QyjWZ_z"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# defining total epochs\n",
        "epochs = 15"
      ],
      "metadata": {
        "id": "cPYNsoyn8h-A"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "**Model 1:**\n",
        "\n",
        "      INPUT => MobileNetV3Large => GAP => DO => SOFTMAX\n",
        "\n",
        "Observations:\n",
        "\n",
        "\n",
        "\n",
        "1.   Trainable parameters: 9,610.\n",
        "2.   Best epoch: 10\n",
        "    *   loss: 0.1080 - accuracy: 0.9818 - val_loss: 0.1134 - val_accuracy: 0.9716\n",
        "3.   From epoch 11 onwards model starts overfitting.\n",
        "\n"
      ],
      "metadata": {
        "id": "kLCf05-lE21w"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# defining model architecture, subsequent training, and plotting graphs\n",
        "# building the model\n",
        "model = mobilenetv3l.MobileNetV3L.build(width=image_width, height=image_height, depth=3,\n",
        "                                        classes=num_classes, data_aug=False, dense_layer=False)\n",
        "\n",
        "# compiling and summarizing the model\n",
        "model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.001),\n",
        "              loss=tf.keras.losses.CategoricalCrossentropy(from_logits=False),\n",
        "              metrics=['accuracy'])\n",
        "print('Model Summary...')\n",
        "model.summary()\n",
        "\n",
        "# defining callback - checkpoint\n",
        "checkpoint = ModelCheckpoint('Models/MobileNetV3L_Basic/mobilenetv3l_basic_{epoch:02d}_{val_loss:.3f}.h5', monitor='val_loss', \n",
        "                             mode='min', verbose=0, save_freq='epoch')\n",
        "\n",
        "# training the model on defined parameters\n",
        "print('\\n\\nModel Training...')\n",
        "history = model.fit(\n",
        "  train_ds, validation_data=val_ds, class_weight=class_weight, epochs=epochs, verbose=1,\n",
        "  batch_size=batch_size, callbacks=[checkpoint] \n",
        ")\n",
        "\n",
        "# plot accuracy and loss graphs of the trained model\n",
        "print('\\n\\nModel Graphs...')\n",
        "fig = helper_functions.plot_graphs(history=history)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "lx5igNZaWaoV",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "921815c8-8157-40de-d2e6-88d8fd517b03"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:tensorflow:Using a while_loop for converting RngReadAndSkip\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting StatelessRandomUniformV2\n",
            "WARNING:tensorflow:Using a while_loop for converting RngReadAndSkip\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting StatelessRandomUniformV2\n",
            "WARNING:tensorflow:Using a while_loop for converting ImageProjectiveTransformV3\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/mobilenet_v3/weights_mobilenet_v3_large_224_1.0_float_no_top_v2.h5\n",
            "12683000/12683000 [==============================] - 2s 0us/step\n",
            "Model Summary...\n",
            "Model: \"MobileNetV3Large\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_2 (InputLayer)        [(None, 224, 224, 3)]     0         \n",
            "                                                                 \n",
            " MobilenetV3large (Functiona  (None, 7, 7, 960)        2996352   \n",
            " l)                                                              \n",
            "                                                                 \n",
            " global_average_pooling2d (G  (None, 960)              0         \n",
            " lobalAveragePooling2D)                                          \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 960)               0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 10)                9610      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,005,962\n",
            "Trainable params: 9,610\n",
            "Non-trainable params: 2,996,352\n",
            "_________________________________________________________________\n",
            "\n",
            "\n",
            "Model Training...\n",
            "Epoch 1/15\n",
            "287/287 [==============================] - 55s 106ms/step - loss: 0.7042 - accuracy: 0.8879 - val_loss: 0.1408 - val_accuracy: 0.9608\n",
            "Epoch 2/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.2460 - accuracy: 0.9627 - val_loss: 0.1196 - val_accuracy: 0.9675\n",
            "Epoch 3/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1924 - accuracy: 0.9698 - val_loss: 0.1161 - val_accuracy: 0.9691\n",
            "Epoch 4/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1670 - accuracy: 0.9729 - val_loss: 0.1084 - val_accuracy: 0.9700\n",
            "Epoch 5/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1527 - accuracy: 0.9761 - val_loss: 0.1102 - val_accuracy: 0.9707\n",
            "Epoch 6/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1400 - accuracy: 0.9785 - val_loss: 0.1092 - val_accuracy: 0.9707\n",
            "Epoch 7/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1278 - accuracy: 0.9793 - val_loss: 0.1076 - val_accuracy: 0.9720\n",
            "Epoch 8/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1215 - accuracy: 0.9806 - val_loss: 0.1139 - val_accuracy: 0.9696\n",
            "Epoch 9/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1124 - accuracy: 0.9817 - val_loss: 0.1111 - val_accuracy: 0.9703\n",
            "Epoch 10/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1080 - accuracy: 0.9818 - val_loss: 0.1134 - val_accuracy: 0.9716\n",
            "Epoch 11/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1029 - accuracy: 0.9822 - val_loss: 0.1118 - val_accuracy: 0.9715\n",
            "Epoch 12/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1012 - accuracy: 0.9828 - val_loss: 0.1163 - val_accuracy: 0.9710\n",
            "Epoch 13/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0950 - accuracy: 0.9838 - val_loss: 0.1168 - val_accuracy: 0.9702\n",
            "Epoch 14/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0863 - accuracy: 0.9848 - val_loss: 0.1214 - val_accuracy: 0.9707\n",
            "Epoch 15/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0936 - accuracy: 0.9840 - val_loss: 0.1238 - val_accuracy: 0.9691\n",
            "\n",
            "\n",
            "Model Graphs...\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**Model 2:**\n",
        "\n",
        "      INPUT > MobileNetV3Large => GAP => FC => DO => SOFTMAX\n",
        "\n",
        "Observations:\n",
        "\n",
        "\n",
        "1.   Trainable parameters: 124,298.\n",
        "2.   Best epoch: 4\n",
        "    *   loss: 0.0939 - accuracy: 0.9842 - val_loss: 0.1100 - val_accuracy: 0.9731\n",
        "3.   Quickly starts overfitting from epoch 5 onwards."
      ],
      "metadata": {
        "id": "-zThgM9WFs72"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# defining model architecture, subsequent training, and plotting graphs\n",
        "# building the model\n",
        "model = mobilenetv3l.MobileNetV3L.build(width=image_width, height=image_height, depth=3,\n",
        "                                        classes=num_classes, data_aug=False, dense_layer=True)\n",
        "\n",
        "# compiling and summarizing the model\n",
        "model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.001),\n",
        "              loss=tf.keras.losses.CategoricalCrossentropy(from_logits=False),\n",
        "              metrics=['accuracy'])\n",
        "print('Model Summary...')\n",
        "model.summary()\n",
        "\n",
        "# defining callback - checkpoint\n",
        "checkpoint = ModelCheckpoint('Models/MobileNetV3L_Dense/mobilenetv3l_dense_{epoch:02d}_{val_loss:.3f}.h5', monitor='val_loss', \n",
        "                             mode='min', verbose=0, save_freq='epoch')\n",
        "\n",
        "# training the model on defined parameters\n",
        "print('\\n\\nModel Training...')\n",
        "history = model.fit(\n",
        "  train_ds, validation_data=val_ds, class_weight=class_weight, epochs=epochs, verbose=1,\n",
        "  batch_size=batch_size, callbacks=[checkpoint] \n",
        ")\n",
        "\n",
        "# plot accuracy and loss graphs of the trained model\n",
        "print('\\n\\nModel Graphs...')\n",
        "fig = helper_functions.plot_graphs(history=history)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "J57W-cXaQIef",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "d346fe2e-56ca-4840-9b63-ed32b48ee88b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model Summary...\n",
            "Model: \"MobileNetV3Large\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_4 (InputLayer)        [(None, 224, 224, 3)]     0         \n",
            "                                                                 \n",
            " MobilenetV3large (Functiona  (None, 7, 7, 960)        2996352   \n",
            " l)                                                              \n",
            "                                                                 \n",
            " global_average_pooling2d_1   (None, 960)              0         \n",
            " (GlobalAveragePooling2D)                                        \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 128)               123008    \n",
            "                                                                 \n",
            " dropout_1 (Dropout)         (None, 128)               0         \n",
            "                                                                 \n",
            " dense_2 (Dense)             (None, 10)                1290      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,120,650\n",
            "Trainable params: 124,298\n",
            "Non-trainable params: 2,996,352\n",
            "_________________________________________________________________\n",
            "\n",
            "\n",
            "Model Training...\n",
            "Epoch 1/15\n",
            "287/287 [==============================] - 32s 101ms/step - loss: 0.4336 - accuracy: 0.9324 - val_loss: 0.1276 - val_accuracy: 0.9626\n",
            "Epoch 2/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1905 - accuracy: 0.9686 - val_loss: 0.1148 - val_accuracy: 0.9678\n",
            "Epoch 3/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.1286 - accuracy: 0.9786 - val_loss: 0.1116 - val_accuracy: 0.9717\n",
            "Epoch 4/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0939 - accuracy: 0.9842 - val_loss: 0.1100 - val_accuracy: 0.9731\n",
            "Epoch 5/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0714 - accuracy: 0.9885 - val_loss: 0.1204 - val_accuracy: 0.9697\n",
            "Epoch 6/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0576 - accuracy: 0.9896 - val_loss: 0.1228 - val_accuracy: 0.9703\n",
            "Epoch 7/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0496 - accuracy: 0.9906 - val_loss: 0.1202 - val_accuracy: 0.9719\n",
            "Epoch 8/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0381 - accuracy: 0.9942 - val_loss: 0.1496 - val_accuracy: 0.9673\n",
            "Epoch 9/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0374 - accuracy: 0.9939 - val_loss: 0.1416 - val_accuracy: 0.9698\n",
            "Epoch 10/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0259 - accuracy: 0.9956 - val_loss: 0.1375 - val_accuracy: 0.9706\n",
            "Epoch 11/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0191 - accuracy: 0.9968 - val_loss: 0.1365 - val_accuracy: 0.9715\n",
            "Epoch 12/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0207 - accuracy: 0.9971 - val_loss: 0.1498 - val_accuracy: 0.9716\n",
            "Epoch 13/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0282 - accuracy: 0.9949 - val_loss: 0.1436 - val_accuracy: 0.9735\n",
            "Epoch 14/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0174 - accuracy: 0.9974 - val_loss: 0.1469 - val_accuracy: 0.9734\n",
            "Epoch 15/15\n",
            "287/287 [==============================] - 28s 98ms/step - loss: 0.0159 - accuracy: 0.9974 - val_loss: 0.1557 - val_accuracy: 0.9702\n",
            "\n",
            "\n",
            "Model Graphs...\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**Model 3:**\n",
        "\n",
        "      INPUT => AUGMENTATION => MobileNetV3Large => GAP => DO => SOFTMAX\n",
        "\n",
        "Observations:\n",
        "\n",
        "\n",
        "1.   Trainable parameters: 9,610.\n",
        "2.   Validation loss is always lower than training loss.\n",
        "    *   Probable Reason: Due to the use of data augmentation technique, the model might be seeing harder examples while training. \n",
        "\n"
      ],
      "metadata": {
        "id": "0ocLbAdqG9gd"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# defining model architecture, subsequent training, and plotting graphs\n",
        "# building the model\n",
        "model = mobilenetv3l.MobileNetV3L.build(width=image_width, height=image_height, depth=3,\n",
        "                                        classes=num_classes, data_aug=True, dense_layer=False)\n",
        "\n",
        "# compiling and summarizing the model\n",
        "model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.001),\n",
        "              loss=tf.keras.losses.CategoricalCrossentropy(from_logits=False),\n",
        "              metrics=['accuracy'])\n",
        "print('Model Summary...')\n",
        "model.summary()\n",
        "\n",
        "# defining callback - checkpoint\n",
        "checkpoint = ModelCheckpoint('Models/MobileNetV3L_Aug/mobilenetv3l_aug_{epoch:02d}_{val_loss:.3f}.h5', monitor='val_loss', \n",
        "                             mode='min', verbose=0, save_freq='epoch')\n",
        "\n",
        "# training the model on defined parameters\n",
        "print('\\n\\nModel Training...')\n",
        "history = model.fit(\n",
        "  train_ds, validation_data=val_ds, class_weight=class_weight, epochs=epochs, verbose=1,\n",
        "  batch_size=batch_size, callbacks=[checkpoint] \n",
        ")\n",
        "\n",
        "# plot accuracy and loss graphs of the trained model\n",
        "print('\\n\\nModel Graphs...')\n",
        "fig = helper_functions.plot_graphs(history=history)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "5H6igs8VQMan",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "283d7fc2-be86-4c7f-d599-d17d5d20d812"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model Summary...\n",
            "Model: \"MobileNetV3Large\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_6 (InputLayer)        [(None, 224, 224, 3)]     0         \n",
            "                                                                 \n",
            " sequential_2 (Sequential)   (None, 224, 224, 3)       0         \n",
            "                                                                 \n",
            " MobilenetV3large (Functiona  (None, 7, 7, 960)        2996352   \n",
            " l)                                                              \n",
            "                                                                 \n",
            " global_average_pooling2d_2   (None, 960)              0         \n",
            " (GlobalAveragePooling2D)                                        \n",
            "                                                                 \n",
            " dropout_2 (Dropout)         (None, 960)               0         \n",
            "                                                                 \n",
            " dense_3 (Dense)             (None, 10)                9610      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,005,962\n",
            "Trainable params: 9,610\n",
            "Non-trainable params: 2,996,352\n",
            "_________________________________________________________________\n",
            "\n",
            "\n",
            "Model Training...\n",
            "Epoch 1/15\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:tensorflow:Using a while_loop for converting RngReadAndSkip\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting StatelessRandomUniformV2\n",
            "WARNING:tensorflow:Using a while_loop for converting RngReadAndSkip\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting StatelessRandomUniformV2\n",
            "WARNING:tensorflow:Using a while_loop for converting ImageProjectiveTransformV3\n",
            "WARNING:tensorflow:Using a while_loop for converting RngReadAndSkip\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting StatelessRandomUniformV2\n",
            "WARNING:tensorflow:Using a while_loop for converting RngReadAndSkip\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting Bitcast\n",
            "WARNING:tensorflow:Using a while_loop for converting StatelessRandomUniformV2\n",
            "WARNING:tensorflow:Using a while_loop for converting ImageProjectiveTransformV3\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "287/287 [==============================] - 107s 355ms/step - loss: 0.8953 - accuracy: 0.8467 - val_loss: 0.1468 - val_accuracy: 0.9570\n",
            "Epoch 2/15\n",
            "287/287 [==============================] - 101s 351ms/step - loss: 0.4078 - accuracy: 0.9311 - val_loss: 0.1182 - val_accuracy: 0.9672\n",
            "Epoch 3/15\n",
            "287/287 [==============================] - 100s 349ms/step - loss: 0.3634 - accuracy: 0.9366 - val_loss: 0.1109 - val_accuracy: 0.9694\n",
            "Epoch 4/15\n",
            "287/287 [==============================] - 100s 350ms/step - loss: 0.3348 - accuracy: 0.9421 - val_loss: 0.1107 - val_accuracy: 0.9693\n",
            "Epoch 5/15\n",
            "287/287 [==============================] - 101s 352ms/step - loss: 0.3145 - accuracy: 0.9448 - val_loss: 0.1130 - val_accuracy: 0.9688\n",
            "Epoch 6/15\n",
            "287/287 [==============================] - 101s 352ms/step - loss: 0.3104 - accuracy: 0.9455 - val_loss: 0.1077 - val_accuracy: 0.9701\n",
            "Epoch 7/15\n",
            "287/287 [==============================] - 101s 352ms/step - loss: 0.2939 - accuracy: 0.9485 - val_loss: 0.1057 - val_accuracy: 0.9710\n",
            "Epoch 8/15\n",
            "287/287 [==============================] - 101s 350ms/step - loss: 0.2819 - accuracy: 0.9496 - val_loss: 0.1145 - val_accuracy: 0.9677\n",
            "Epoch 9/15\n",
            "287/287 [==============================] - 101s 350ms/step - loss: 0.2722 - accuracy: 0.9519 - val_loss: 0.1061 - val_accuracy: 0.9719\n",
            "Epoch 10/15\n",
            "287/287 [==============================] - 101s 350ms/step - loss: 0.2732 - accuracy: 0.9511 - val_loss: 0.1063 - val_accuracy: 0.9721\n",
            "Epoch 11/15\n",
            "287/287 [==============================] - 100s 350ms/step - loss: 0.2610 - accuracy: 0.9548 - val_loss: 0.1087 - val_accuracy: 0.9717\n",
            "Epoch 12/15\n",
            "287/287 [==============================] - 101s 352ms/step - loss: 0.2696 - accuracy: 0.9518 - val_loss: 0.1116 - val_accuracy: 0.9710\n",
            "Epoch 13/15\n",
            "287/287 [==============================] - 101s 350ms/step - loss: 0.2674 - accuracy: 0.9539 - val_loss: 0.1094 - val_accuracy: 0.9719\n",
            "Epoch 14/15\n",
            "287/287 [==============================] - 101s 351ms/step - loss: 0.2707 - accuracy: 0.9539 - val_loss: 0.1130 - val_accuracy: 0.9697\n",
            "Epoch 15/15\n",
            "287/287 [==============================] - 100s 350ms/step - loss: 0.2741 - accuracy: 0.9531 - val_loss: 0.1106 - val_accuracy: 0.9714\n",
            "\n",
            "\n",
            "Model Graphs...\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x576 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**Conclusion:**\n",
        "\n",
        "\n",
        "\n",
        "1.   Best epoch of Model 2 performs slightly better than best epoch of Model 1, however, trainable parameters of model 2 are approximately 12 times higher than that of Model 1.\n",
        "2.   Data augmentation technique can result in lower validation loss than training loss.\n",
        "\n"
      ],
      "metadata": {
        "id": "CLXL-eq9OZx4"
      }
    }
  ]
}