6_training_resnet50v2.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 ResNet50V2 model."
      ],
      "metadata": {
        "id": "wVxuJ9U-U0nH"
      }
    },
    {
      "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 resnet50v2 # 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": {
        "id": "iVVzOCdjQFi2",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1f4e63a5-9a41-4edf-c76b-cffb4ca75fb1"
      },
      "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": "50231a8c-f4da-44a3-9575-5d0eebe63759"
      },
      "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": "2bbc3b8c-9639-450f-9760-e135c6d660b5"
      },
      "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": "markdown",
      "source": [
        "**Model:**\n",
        "\n",
        "      INPUT => ResNet50V2 => GAP => DO => SOFTMAX\n",
        "\n",
        "Observations:\n",
        "\n",
        "\n",
        "\n",
        "1.   Trainable parameters: 20,490.\n",
        "2.   Best epoch: 6.\n",
        "    *   loss: 0.1240 - accuracy: 0.9780 - val_loss: 0.1517 - val_accuracy: 0.9593\n",
        "3.   From epoch 7 onwards model starts overfitting quickly.\n",
        "\n",
        "      "
      ],
      "metadata": {
        "id": "2_Hi83XlU8yG"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# defining model architecture, subsequent training, and plotting graphs\n",
        "# building the model\n",
        "model = resnet50v2.ResNet50_V2.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/ResNet50V2_Basic/resnet50v2_basic_{epoch:02d}_{val_loss:.3f}.h5', monitor='val_loss', \n",
        "                             mode='min', verbose=0, save_freq='epoch')\n",
        "# total epochs\n",
        "epochs = 15 \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": "a1f3501e-d854-443e-b2c3-b5748c534417"
      },
      "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/resnet/resnet50v2_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
            "94668760/94668760 [==============================] - 3s 0us/step\n",
            "Model Summary...\n",
            "Model: \"ResNet50V2\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_2 (InputLayer)        [(None, 224, 224, 3)]     0         \n",
            "                                                                 \n",
            " tf.math.truediv (TFOpLambda  (None, 224, 224, 3)      0         \n",
            " )                                                               \n",
            "                                                                 \n",
            " tf.math.subtract (TFOpLambd  (None, 224, 224, 3)      0         \n",
            " a)                                                              \n",
            "                                                                 \n",
            " resnet50v2 (Functional)     (None, 7, 7, 2048)        23564800  \n",
            "                                                                 \n",
            " global_average_pooling2d (G  (None, 2048)             0         \n",
            " lobalAveragePooling2D)                                          \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 2048)              0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 10)                20490     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 23,585,290\n",
            "Trainable params: 20,490\n",
            "Non-trainable params: 23,564,800\n",
            "_________________________________________________________________\n",
            "\n",
            "\n",
            "Model Training...\n",
            "Epoch 1/15\n",
            "287/287 [==============================] - 93s 238ms/step - loss: 0.6092 - accuracy: 0.9022 - val_loss: 0.1473 - val_accuracy: 0.9576\n",
            "Epoch 2/15\n",
            "287/287 [==============================] - 66s 231ms/step - loss: 0.2625 - accuracy: 0.9575 - val_loss: 0.1525 - val_accuracy: 0.9571\n",
            "Epoch 3/15\n",
            "287/287 [==============================] - 67s 234ms/step - loss: 0.1970 - accuracy: 0.9667 - val_loss: 0.1416 - val_accuracy: 0.9609\n",
            "Epoch 4/15\n",
            "287/287 [==============================] - 68s 238ms/step - loss: 0.1694 - accuracy: 0.9718 - val_loss: 0.1392 - val_accuracy: 0.9607\n",
            "Epoch 5/15\n",
            "287/287 [==============================] - 69s 241ms/step - loss: 0.1498 - accuracy: 0.9742 - val_loss: 0.1403 - val_accuracy: 0.9619\n",
            "Epoch 6/15\n",
            "287/287 [==============================] - 69s 240ms/step - loss: 0.1240 - accuracy: 0.9780 - val_loss: 0.1517 - val_accuracy: 0.9593\n",
            "Epoch 7/15\n",
            "287/287 [==============================] - 69s 242ms/step - loss: 0.1148 - accuracy: 0.9792 - val_loss: 0.1507 - val_accuracy: 0.9612\n",
            "Epoch 8/15\n",
            "287/287 [==============================] - 70s 243ms/step - loss: 0.1085 - accuracy: 0.9805 - val_loss: 0.1556 - val_accuracy: 0.9605\n",
            "Epoch 9/15\n",
            "287/287 [==============================] - 70s 242ms/step - loss: 0.1012 - accuracy: 0.9822 - val_loss: 0.1556 - val_accuracy: 0.9618\n",
            "Epoch 10/15\n",
            "287/287 [==============================] - 70s 243ms/step - loss: 0.0987 - accuracy: 0.9832 - val_loss: 0.1628 - val_accuracy: 0.9608\n",
            "Epoch 11/15\n",
            "287/287 [==============================] - 70s 243ms/step - loss: 0.0855 - accuracy: 0.9837 - val_loss: 0.1615 - val_accuracy: 0.9619\n",
            "Epoch 12/15\n",
            "287/287 [==============================] - 70s 243ms/step - loss: 0.0877 - accuracy: 0.9842 - val_loss: 0.1670 - val_accuracy: 0.9599\n",
            "Epoch 13/15\n",
            "287/287 [==============================] - 70s 243ms/step - loss: 0.0782 - accuracy: 0.9854 - val_loss: 0.1701 - val_accuracy: 0.9614\n",
            "Epoch 14/15\n",
            "287/287 [==============================] - 70s 244ms/step - loss: 0.0771 - accuracy: 0.9855 - val_loss: 0.1768 - val_accuracy: 0.9587\n",
            "Epoch 15/15\n",
            "287/287 [==============================] - 70s 243ms/step - loss: 0.0803 - accuracy: 0.9846 - val_loss: 0.1937 - val_accuracy: 0.9586\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.   Epoch 6 of ResNet50V2 model gives the best results. However, the results are not as good as given by best model from MobileNetV3Large. Also, the number of trainable parameters are 2X more than that of its counterpart.\n",
        "2.   Size of ResNet50V2 is approximately 90 MB, which is approximately more than 8 times that of MobileNetV3Large (approximately 12 MB).\n"
      ],
      "metadata": {
        "id": "IGgwmrnPVnEi"
      }
    }
  ]
}