3. Quantization_Aware_Training .ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "Quantization_Aware_Training.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "wgXcu1c4prME",
        "outputId": "2c8061a7-c3f9-4c72-f5a9-21ab385f8d64"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "MWkwmc1UbSlW",
        "outputId": "088e7e5a-8fb9-49d8-834a-70a4159eacf2"
      },
      "source": [
        "!nvidia-smi"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Sun May 30 19:03:13 2021       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 465.19.01    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 T4            Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   54C    P8    10W /  70W |      0MiB / 15109MiB |      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"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "ARmFWwk9qLyk",
        "outputId": "b0be504b-e663-48f8-a52e-e5955e79d1d7"
      },
      "source": [
        "ls"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\u001b[0m\u001b[01;34mdrive\u001b[0m/  \u001b[01;34msample_data\u001b[0m/\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "gMyr2dC4qM8M",
        "outputId": "3bc906f4-abbe-4f45-f1ae-6964a574c297"
      },
      "source": [
        "cd drive/MyDrive/"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/drive/MyDrive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "TFGvEnHWqO-h",
        "outputId": "1840b897-119a-4aca-cf39-8971064ba65c"
      },
      "source": [
        "ls"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            " 0002cc93b.jpg.txt                  \u001b[0m\u001b[01;34mmalikbhai\u001b[0m/   \u001b[01;34mTRT-OP\u001b[0m/\n",
            "\u001b[01;34m'Codebugged Official Repository'\u001b[0m/   \u001b[01;34mMRV\u001b[0m/         \u001b[01;34mvideo_class\u001b[0m/\n",
            "\u001b[01;34m'Colab Notebooks'\u001b[0m/                  \u001b[01;34mNVIDIA\u001b[0m/      \u001b[01;34mYOLO\u001b[0m/\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gZ7mc1oCqQek"
      },
      "source": [
        "mkdir QAT"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "eBrT3kD_qToU",
        "outputId": "9acf70fd-e69c-4192-e729-6ca711c9b152"
      },
      "source": [
        "cd QAT"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/drive/MyDrive/QAT\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nXb-CReC1iw7"
      },
      "source": [
        "## Installing Necessary Libraries\n",
        "\n",
        "### TF-Nightly - A version of tensorflow resolving build failure"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "gjc81l8_bc1J",
        "outputId": "aeaf04a2-3900-4589-d3b2-60e94b8be12a"
      },
      "source": [
        "!pip uninstall -y tensorflow\n",
        "!pip install -q tf-nightly\n",
        "!pip install -qq tensorflow-model-optimization"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Uninstalling tensorflow-2.5.0:\n",
            "  Successfully uninstalled tensorflow-2.5.0\n",
            "\u001b[K     |████████████████████████████████| 454.4MB 40kB/s \n",
            "\u001b[K     |████████████████████████████████| 5.9MB 20.5MB/s \n",
            "\u001b[K     |████████████████████████████████| 1.3MB 7.9MB/s \n",
            "\u001b[K     |████████████████████████████████| 4.2MB 38.7MB/s \n",
            "\u001b[K     |████████████████████████████████| 471kB 49.1MB/s \n",
            "\u001b[K     |████████████████████████████████| 174kB 9.1MB/s \n",
            "\u001b[?25h"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "JqBkIqcAdNhF",
        "outputId": "91a62b66-f28b-4bc6-bc6b-ea9ca3c7a3bb"
      },
      "source": [
        "#Data Gathering from Google API \n",
        "!wget https://storage.googleapis.com/laurencemoroney-blog.appspot.com/rps.zip \n",
        "!wget https://storage.googleapis.com/laurencemoroney-blog.appspot.com/rps-test-set.zip"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-05-30 19:05:47--  https://storage.googleapis.com/laurencemoroney-blog.appspot.com/rps.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 74.125.20.128, 108.177.98.128, 74.125.197.128, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|74.125.20.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 200682221 (191M) [application/zip]\n",
            "Saving to: ‘rps.zip’\n",
            "\n",
            "rps.zip             100%[===================>] 191.38M  61.4MB/s    in 3.1s    \n",
            "\n",
            "2021-05-30 19:05:50 (61.4 MB/s) - ‘rps.zip’ saved [200682221/200682221]\n",
            "\n",
            "--2021-05-30 19:05:50--  https://storage.googleapis.com/laurencemoroney-blog.appspot.com/rps-test-set.zip\n",
            "Resolving storage.googleapis.com (storage.googleapis.com)... 74.125.199.128, 74.125.142.128, 74.125.195.128, ...\n",
            "Connecting to storage.googleapis.com (storage.googleapis.com)|74.125.199.128|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 29516758 (28M) [application/zip]\n",
            "Saving to: ‘rps-test-set.zip’\n",
            "\n",
            "rps-test-set.zip    100%[===================>]  28.15M  58.7MB/s    in 0.5s    \n",
            "\n",
            "2021-05-30 19:05:51 (58.7 MB/s) - ‘rps-test-set.zip’ saved [29516758/29516758]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MSJZ_DuEdQfg"
      },
      "source": [
        "!unzip -q rps.zip\n",
        "!unzip -qq rps-test-set.zip"
      ],
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "9eQ9UDv0dSr_",
        "outputId": "6009e892-5e32-404f-855c-03fe254dc562"
      },
      "source": [
        "from imutils import paths\n",
        "from pprint import pprint\n",
        "from collections import Counter\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import *\n",
        "from tensorflow.keras.utils import plot_model\n",
        "import tensorflow_model_optimization as tfmot\n",
        "import re \n",
        "\n",
        "import os\n",
        "from sys import getsizeof\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import matplotlib.image as mpimg\n",
        "import seaborn as sns\n",
        "from sklearn.metrics import accuracy_score\n",
        "from sys import getsizeof\n",
        "\n",
        "import numpy as np\n",
        "np.random.seed(42)\n",
        "\n",
        "\n",
        "print(tf.__version__)"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2.6.0-dev20210530\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MrwKNJPQdVER"
      },
      "source": [
        "def get_file_size(file_path):\n",
        "    size = os.path.getsize(file_path)\n",
        "    return size\n",
        "\n",
        "def convert_bytes(size, unit=None):\n",
        "    if unit == \"KB\":\n",
        "        return print('File Size: ' + str(round(size/1024, 3)) + 'Kilobytes')\n",
        "    elif unit == 'MB':\n",
        "        return print('File Size: ' + str(round(size/(1024*1024), 3)) + 'Megabytes')\n",
        "    else:\n",
        "        return print('File Size: ' + str(size) + 'bytes')"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_uie-8xtdXqD"
      },
      "source": [
        "train_dir='rps/'\n",
        "test_dir='rps-test-set/'"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "mqlSW5zJdask",
        "outputId": "6d5bfc06-5500-4a0d-9b70-2942b2beaac5"
      },
      "source": [
        "image_paths = list(paths.list_images(\"rps\"))\n",
        "np.random.shuffle(image_paths)\n",
        "image_paths[:5]"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['rps/paper/paper02-076.png',\n",
              " 'rps/rock/rock02-026.png',\n",
              " 'rps/rock/rock03-010.png',\n",
              " 'rps/paper/paper04-027.png',\n",
              " 'rps/scissors/testscissors01-061.png']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "wEtD1Gu6dc6_",
        "outputId": "c7eb3581-c7c7-4fec-e971-d9d30340da86"
      },
      "source": [
        "#Counting images per class\n",
        "labels = []\n",
        "for image_path in image_paths:\n",
        "    label = image_path.split(\"/\")[1]\n",
        "    labels.append(label)\n",
        "class_count = Counter(labels) \n",
        "pprint(class_count)"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Counter({'paper': 840, 'rock': 840, 'scissors': 840})\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u4GMV8oNdfCA"
      },
      "source": [
        "#Defining Hyperparameters\n",
        "BATCH_SIZE = 128\n",
        "EPOCHS = 20\n",
        "IMG_SIZE = 150\n",
        "NUM_CLASSES=3"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "H5LntKRldgrr",
        "outputId": "37901aaf-f690-48ca-eb68-118d04f0e9af"
      },
      "source": [
        "training_datagen = tf.keras.preprocessing.image.ImageDataGenerator(\n",
        "      rescale = 1./255,\n",
        "\t    rotation_range=40,\n",
        "      width_shift_range=0.2,\n",
        "      height_shift_range=0.2,\n",
        "      shear_range=0.2,\n",
        "      zoom_range=0.2,\n",
        "      horizontal_flip=True,\n",
        "      vertical_flip=True,\n",
        "      fill_mode='nearest')\n",
        "\n",
        "\n",
        "validation_datagen = tf.keras.preprocessing.image.ImageDataGenerator(rescale = 1./255)\n",
        "\n",
        "train_generator = training_datagen.flow_from_directory(\n",
        "\ttrain_dir,\n",
        "\ttarget_size=(IMG_SIZE,IMG_SIZE),\n",
        "\tclass_mode='categorical',\n",
        "    batch_size=BATCH_SIZE\n",
        ")\n",
        "\n",
        "validation_generator = validation_datagen.flow_from_directory(\n",
        "\ttest_dir,\n",
        "\ttarget_size=(IMG_SIZE,IMG_SIZE),\n",
        "\tclass_mode='categorical',\n",
        "    batch_size=BATCH_SIZE\n",
        ")"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 2520 images belonging to 3 classes.\n",
            "Found 372 images belonging to 3 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "hysiXIqpdifC",
        "outputId": "1a10b45c-1d6a-4053-8aaf-31a88a0357cb"
      },
      "source": [
        "train_generator.class_indices"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'paper': 0, 'rock': 1, 'scissors': 2}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nISiM7Ildk5H"
      },
      "source": [
        "#Defining the model\n",
        "def training_model():\n",
        "    model = Sequential([\n",
        "        tf.keras.layers.Conv2D(64, (3,3), activation='relu', input_shape=(150, 150, 3)),\n",
        "        tf.keras.layers.MaxPooling2D(2, 2),\n",
        "        # The second convolution\n",
        "        tf.keras.layers.Conv2D(64, (3,3), activation='relu'),\n",
        "        tf.keras.layers.MaxPooling2D(2,2),\n",
        "        # The third convolution\n",
        "        tf.keras.layers.Conv2D(128, (3,3), activation='relu'),\n",
        "        tf.keras.layers.MaxPooling2D(2,2),\n",
        "        # The fourth convolution\n",
        "        tf.keras.layers.Conv2D(128, (3,3), activation='relu'),\n",
        "        tf.keras.layers.MaxPooling2D(2,2),\n",
        "        # Flatten the results to feed into a DNN\n",
        "        tf.keras.layers.Flatten(),\n",
        "        tf.keras.layers.Dropout(0.5),\n",
        "        # 512 neuron hidden layer\n",
        "        tf.keras.layers.Dense(512, activation='relu'),\n",
        "        tf.keras.layers.Dense(NUM_CLASSES, activation='softmax')                       \n",
        "    ])\n",
        "    return model"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "23dvT4RLdmfP",
        "outputId": "f1cbe943-11ca-4b17-a332-eceef43ccc27"
      },
      "source": [
        "model = training_model()\n",
        "model.summary()"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 148, 148, 64)      1792      \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 74, 74, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 72, 72, 64)        36928     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 36, 36, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 34, 34, 128)       73856     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 17, 17, 128)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 15, 15, 128)       147584    \n",
            "_________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2 (None, 7, 7, 128)         0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 6272)              0         \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 6272)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 512)               3211776   \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 3)                 1539      \n",
            "=================================================================\n",
            "Total params: 3,473,475\n",
            "Trainable params: 3,473,475\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OeGUBTRrB0Yg"
      },
      "source": [
        "### Compiling and training the Baseline Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "sisgfz4NdoBa",
        "outputId": "b66b6313-5992-4e17-fdfa-9ebd66e5edae"
      },
      "source": [
        "#Compile the Model\n",
        "model.compile(loss = 'categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])\n",
        "\n",
        "#Train the Model\n",
        "history = model.fit(train_generator, epochs=EPOCHS, validation_data = validation_generator, validation_steps=3, batch_size=BATCH_SIZE)"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "20/20 [==============================] - 158s 8s/step - loss: 1.3244 - accuracy: 0.3381 - val_loss: 1.0780 - val_accuracy: 0.4839\n",
            "Epoch 2/20\n",
            "20/20 [==============================] - 154s 8s/step - loss: 1.1145 - accuracy: 0.3726 - val_loss: 1.0640 - val_accuracy: 0.4113\n",
            "Epoch 3/20\n",
            "20/20 [==============================] - 150s 7s/step - loss: 1.1020 - accuracy: 0.4071 - val_loss: 1.0900 - val_accuracy: 0.3387\n",
            "Epoch 4/20\n",
            "20/20 [==============================] - 149s 7s/step - loss: 1.0655 - accuracy: 0.4389 - val_loss: 1.0041 - val_accuracy: 0.5645\n",
            "Epoch 5/20\n",
            "20/20 [==============================] - 149s 7s/step - loss: 1.0456 - accuracy: 0.4901 - val_loss: 1.0005 - val_accuracy: 0.5081\n",
            "Epoch 6/20\n",
            "20/20 [==============================] - 148s 7s/step - loss: 0.9125 - accuracy: 0.5611 - val_loss: 0.5844 - val_accuracy: 0.6989\n",
            "Epoch 7/20\n",
            "20/20 [==============================] - 148s 7s/step - loss: 0.9639 - accuracy: 0.5770 - val_loss: 0.7455 - val_accuracy: 0.5860\n",
            "Epoch 8/20\n",
            "20/20 [==============================] - 148s 7s/step - loss: 0.8054 - accuracy: 0.6409 - val_loss: 0.4021 - val_accuracy: 0.9651\n",
            "Epoch 9/20\n",
            "20/20 [==============================] - 148s 7s/step - loss: 0.6702 - accuracy: 0.7258 - val_loss: 0.2934 - val_accuracy: 0.9651\n",
            "Epoch 10/20\n",
            "20/20 [==============================] - 148s 7s/step - loss: 0.6086 - accuracy: 0.7381 - val_loss: 0.1771 - val_accuracy: 0.9516\n",
            "Epoch 11/20\n",
            "20/20 [==============================] - 147s 7s/step - loss: 0.7552 - accuracy: 0.7194 - val_loss: 0.4361 - val_accuracy: 0.7258\n",
            "Epoch 12/20\n",
            "20/20 [==============================] - 147s 7s/step - loss: 0.4546 - accuracy: 0.8063 - val_loss: 0.1490 - val_accuracy: 0.9731\n",
            "Epoch 13/20\n",
            "20/20 [==============================] - 147s 7s/step - loss: 0.4805 - accuracy: 0.8194 - val_loss: 0.1141 - val_accuracy: 0.9758\n",
            "Epoch 14/20\n",
            "20/20 [==============================] - 147s 7s/step - loss: 0.3623 - accuracy: 0.8615 - val_loss: 0.0503 - val_accuracy: 0.9839\n",
            "Epoch 15/20\n",
            "20/20 [==============================] - 148s 7s/step - loss: 0.3960 - accuracy: 0.8516 - val_loss: 0.1676 - val_accuracy: 0.9731\n",
            "Epoch 16/20\n",
            "20/20 [==============================] - 151s 8s/step - loss: 0.2699 - accuracy: 0.8933 - val_loss: 0.0423 - val_accuracy: 0.9866\n",
            "Epoch 17/20\n",
            "20/20 [==============================] - 149s 7s/step - loss: 0.2732 - accuracy: 0.8913 - val_loss: 0.0666 - val_accuracy: 0.9785\n",
            "Epoch 18/20\n",
            "20/20 [==============================] - 150s 7s/step - loss: 0.2047 - accuracy: 0.9187 - val_loss: 0.1449 - val_accuracy: 0.9409\n",
            "Epoch 19/20\n",
            "20/20 [==============================] - 152s 8s/step - loss: 0.3048 - accuracy: 0.8956 - val_loss: 0.0242 - val_accuracy: 0.9866\n",
            "Epoch 20/20\n",
            "20/20 [==============================] - 153s 8s/step - loss: 0.2179 - accuracy: 0.9214 - val_loss: 0.1040 - val_accuracy: 0.9731\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6bQ-2yc3dqCX",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "outputId": "88871f9f-d04c-48b4-ec25-f9309530c6de"
      },
      "source": [
        "#Evaluating BaseLine model\n",
        "test_loss, test_acc = model.evaluate(validation_generator, verbose=1)\n",
        "print('Test Accuracy:', test_acc)\n",
        "print('Test Loss:', test_loss)"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "3/3 [==============================] - 6s 2s/step - loss: 0.1040 - accuracy: 0.9731\n",
            "Test Accuracy: 0.9731183052062988\n",
            "Test Loss: 0.10402872413396835\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Cfl6PuykCBSy"
      },
      "source": [
        "###  Plot the Acc and Loss Curve"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u5WssR8tduTC",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "outputId": "88995db1-47b2-4642-dd0d-a786bd8c425f"
      },
      "source": [
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.legend(['train', 'test'], loc='lower right')\n",
        "plt.title('accuracy plot - train vs test')\n",
        "plt.xlabel('epoch')\n",
        "plt.ylabel('accuracy')\n",
        "\n",
        "plt.show()\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.legend(['training loss', 'validation loss'], loc = 'upper right')\n",
        "plt.title('loss plot - training vs validation')\n",
        "plt.xlabel('epoch')\n",
        "plt.ylabel('loss')\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qMYEZjqbCHSF"
      },
      "source": [
        "### Saving our Baseline Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "nW7vmg-udvlW",
        "outputId": "9e9e8810-734c-48b4-c6ef-4390f2d2496f"
      },
      "source": [
        "model.save('rps_model') #for making tflite\n",
        "model.save('rps.h5')"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: rps_model/assets\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "trIIR67YdxPF",
        "outputId": "a36dafb8-cb83-4767-bccf-bebcaae0de66"
      },
      "source": [
        "!du -lh rps_model"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "27M\trps_model/variables\n",
            "4.0K\trps_model/assets\n",
            "27M\trps_model\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "yqnDJWvjdy3R",
        "outputId": "5c3c8422-d258-40ce-cc50-53ad4bda4f7a"
      },
      "source": [
        "convert_bytes(get_file_size('rps.h5'), \"MB\")"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "File Size: 26.557Megabytes\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "ZnHclVShd0To",
        "outputId": "1656cf32-6e44-4d41-aced-a51f6558c194"
      },
      "source": [
        "!tar cvf rps.tar.gz rps_model rps.h5"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "rps_model/\n",
            "rps_model/variables/\n",
            "rps_model/variables/variables.data-00000-of-00001\n",
            "rps_model/variables/variables.index\n",
            "rps_model/assets/\n",
            "rps_model/saved_model.pb\n",
            "rps_model/keras_metadata.pb\n",
            "rps.h5\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TlBvzvW8CW_q"
      },
      "source": [
        "### Quantizing our model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "aapI0QL_d12b",
        "outputId": "436b3362-a80f-499b-8499-aaae820584da"
      },
      "source": [
        "quantize_model = tfmot.quantization.keras.quantize_model\n",
        "\n",
        "# q_aware stands for for quantization aware.\n",
        "q_aware_model = quantize_model(model)\n",
        "\n",
        "# `quantize_model` requires a recompile.\n",
        "q_aware_model.compile(optimizer='rmsprop',\n",
        "              loss='categorical_crossentropy',\n",
        "              metrics=['accuracy'])\n",
        "\n",
        "q_aware_model.summary()"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "quantize_layer (QuantizeLaye (None, 150, 150, 3)       3         \n",
            "_________________________________________________________________\n",
            "quant_conv2d (QuantizeWrappe (None, 148, 148, 64)      1923      \n",
            "_________________________________________________________________\n",
            "quant_max_pooling2d (Quantiz (None, 74, 74, 64)        1         \n",
            "_________________________________________________________________\n",
            "quant_conv2d_1 (QuantizeWrap (None, 72, 72, 64)        37059     \n",
            "_________________________________________________________________\n",
            "quant_max_pooling2d_1 (Quant (None, 36, 36, 64)        1         \n",
            "_________________________________________________________________\n",
            "quant_conv2d_2 (QuantizeWrap (None, 34, 34, 128)       74115     \n",
            "_________________________________________________________________\n",
            "quant_max_pooling2d_2 (Quant (None, 17, 17, 128)       1         \n",
            "_________________________________________________________________\n",
            "quant_conv2d_3 (QuantizeWrap (None, 15, 15, 128)       147843    \n",
            "_________________________________________________________________\n",
            "quant_max_pooling2d_3 (Quant (None, 7, 7, 128)         1         \n",
            "_________________________________________________________________\n",
            "quant_flatten (QuantizeWrapp (None, 6272)              1         \n",
            "_________________________________________________________________\n",
            "quant_dropout (QuantizeWrapp (None, 6272)              1         \n",
            "_________________________________________________________________\n",
            "quant_dense (QuantizeWrapper (None, 512)               3211781   \n",
            "_________________________________________________________________\n",
            "quant_dense_1 (QuantizeWrapp (None, 3)                 1544      \n",
            "=================================================================\n",
            "Total params: 3,474,274\n",
            "Trainable params: 3,473,475\n",
            "Non-trainable params: 799\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "ljzhXGl6d3w3",
        "outputId": "bcf79ff0-1e1e-4da9-c9f7-d2af3b68dc6c"
      },
      "source": [
        "# Training the quantized model\n",
        "q_aware_model.fit(train_generator, epochs=20, validation_data = validation_generator, batch_size=256)"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "20/20 [==============================] - 178s 9s/step - loss: 0.4357 - accuracy: 0.8786 - val_loss: 0.1036 - val_accuracy: 0.9731\n",
            "Epoch 2/20\n",
            "20/20 [==============================] - 185s 9s/step - loss: 0.2417 - accuracy: 0.9175 - val_loss: 0.0440 - val_accuracy: 0.9866\n",
            "Epoch 3/20\n",
            "20/20 [==============================] - 174s 9s/step - loss: 0.2744 - accuracy: 0.9052 - val_loss: 0.0774 - val_accuracy: 0.9677\n",
            "Epoch 4/20\n",
            "20/20 [==============================] - 171s 9s/step - loss: 0.1147 - accuracy: 0.9627 - val_loss: 0.0963 - val_accuracy: 0.9704\n",
            "Epoch 5/20\n",
            "20/20 [==============================] - 171s 8s/step - loss: 0.1755 - accuracy: 0.9417 - val_loss: 0.1835 - val_accuracy: 0.9758\n",
            "Epoch 6/20\n",
            "20/20 [==============================] - 171s 8s/step - loss: 0.1411 - accuracy: 0.9480 - val_loss: 0.0595 - val_accuracy: 0.9839\n",
            "Epoch 7/20\n",
            "20/20 [==============================] - 170s 8s/step - loss: 0.1246 - accuracy: 0.9548 - val_loss: 0.3092 - val_accuracy: 0.8952\n",
            "Epoch 8/20\n",
            "20/20 [==============================] - 170s 8s/step - loss: 0.1621 - accuracy: 0.9417 - val_loss: 0.0998 - val_accuracy: 0.9731\n",
            "Epoch 9/20\n",
            "20/20 [==============================] - 171s 8s/step - loss: 0.2174 - accuracy: 0.9194 - val_loss: 0.1612 - val_accuracy: 0.9489\n",
            "Epoch 10/20\n",
            "20/20 [==============================] - 169s 8s/step - loss: 0.0897 - accuracy: 0.9690 - val_loss: 0.1511 - val_accuracy: 0.9462\n",
            "Epoch 11/20\n",
            "20/20 [==============================] - 168s 8s/step - loss: 0.1304 - accuracy: 0.9556 - val_loss: 0.1916 - val_accuracy: 0.9462\n",
            "Epoch 12/20\n",
            "20/20 [==============================] - 168s 8s/step - loss: 0.1219 - accuracy: 0.9556 - val_loss: 0.1023 - val_accuracy: 0.9597\n",
            "Epoch 13/20\n",
            "20/20 [==============================] - 168s 8s/step - loss: 0.1899 - accuracy: 0.9325 - val_loss: 0.0963 - val_accuracy: 0.9570\n",
            "Epoch 14/20\n",
            "20/20 [==============================] - 173s 9s/step - loss: 0.0752 - accuracy: 0.9762 - val_loss: 0.1189 - val_accuracy: 0.9301\n",
            "Epoch 15/20\n",
            "20/20 [==============================] - 172s 9s/step - loss: 0.1789 - accuracy: 0.9353 - val_loss: 0.0468 - val_accuracy: 0.9758\n",
            "Epoch 16/20\n",
            "20/20 [==============================] - 170s 8s/step - loss: 0.0572 - accuracy: 0.9802 - val_loss: 0.1022 - val_accuracy: 0.9570\n",
            "Epoch 17/20\n",
            "20/20 [==============================] - 170s 8s/step - loss: 0.1327 - accuracy: 0.9599 - val_loss: 0.0327 - val_accuracy: 0.9839\n",
            "Epoch 18/20\n",
            "20/20 [==============================] - 170s 8s/step - loss: 0.0973 - accuracy: 0.9623 - val_loss: 0.0716 - val_accuracy: 0.9731\n",
            "Epoch 19/20\n",
            "20/20 [==============================] - 169s 8s/step - loss: 0.0689 - accuracy: 0.9810 - val_loss: 0.0631 - val_accuracy: 0.9785\n",
            "Epoch 20/20\n",
            "20/20 [==============================] - 169s 8s/step - loss: 0.1025 - accuracy: 0.9639 - val_loss: 0.1229 - val_accuracy: 0.9570\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f8a8003e810>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MtqJjwctd5j7"
      },
      "source": [
        "#Evaluating the quantized model\n",
        "q_aware_model_loss, q_aware_model_accuracy = q_aware_model.evaluate(validation_generator, verbose=1)\n",
        "print('Quant Test Accuracy:', q_aware_model_accuracy)\n",
        "print('Quant Test Loss:', q_aware_model_loss)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p7ZWBnBdd7WI"
      },
      "source": [
        "#Convert the Baseline Model to TF-Lite\n",
        "TF_LITE_MODEL_FILE_NAME = 'tf_lite_model_rps.tflite'\n",
        "\n",
        "tf_lite_converter = tf.lite.TFLiteConverter.from_keras_model(model)\n",
        "tflite_model = tf_lite_converter.convert()\n",
        "\n",
        "tflite_model_name = TF_LITE_MODEL_FILE_NAME\n",
        "open(tflite_model_name, \"wb\").write(tflite_model)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xBgjHeuDC78l"
      },
      "source": [
        "#Convert the Quantise Model to TF-Lite\n",
        "TF_LITE_QUANT_MODEL_FILE_NAME = 'tf_lite_quant_model_rps.tflite'\n",
        "\n",
        "tf_lite_converter = tf.lite.TFLiteConverter.from_keras_model(q_aware_model)\n",
        "tf_lite_converter.optimizations = [tf.lite.Optimize.DEFAULT]\n",
        "tflite_model = tf_lite_converter.convert()\n",
        "\n",
        "tflite_model_name = TF_LITE_QUANT_MODEL_FILE_NAME\n",
        "open(tflite_model_name, \"wb\").write(tflite_model)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gSqEkgbNd98A"
      },
      "source": [
        "#Calculating the File Size\n",
        "convert_bytes(get_file_size('tf_lite_model_rps.tflite'), \"MB\")\n",
        "convert_bytes(get_file_size('tf_lite_quant_model_rps.tflite'), \"MB\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "L0G7Tm75d_ac"
      },
      "source": [
        "images, labels = (next(iter(validation_generator)))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_qfyPM5WeAts"
      },
      "source": [
        "print(images.shape)\n",
        "print(images.dtype)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "f6jXCgMleB_q"
      },
      "source": [
        "#Checking the tensor shape of Baseline model\n",
        "interpreter = tf.lite.Interpreter(model_path = TF_LITE_MODEL_FILE_NAME)\n",
        "input_details = interpreter.get_input_details()\n",
        "output_details = interpreter.get_output_details()\n",
        "\n",
        "print(\"Input Shape:\", input_details[0]['shape'])\n",
        "print(\"Input Type:\", input_details[0]['dtype'])\n",
        "print(\"Output Shape:\", output_details[0]['shape'])\n",
        "print(\"Output Type:\", output_details[0]['dtype'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I4LBN8ileDPT"
      },
      "source": [
        "#Resize the Tensor Shape - Baseline Model\n",
        "interpreter.resize_tensor_input(input_details[0]['index'], (128, 150, 150, 3))\n",
        "interpreter.resize_tensor_input(output_details[0]['index'], (128, 3))\n",
        "interpreter.allocate_tensors()\n",
        "\n",
        "input_details = interpreter.get_input_details()\n",
        "output_details = interpreter.get_output_details()\n",
        "\n",
        "print(\"Input Shape:\", input_details[0]['shape'])\n",
        "print(\"Input Type:\", input_details[0]['dtype'])\n",
        "print(\"Output Shape:\", output_details[0]['shape'])\n",
        "print(\"Output Type:\", output_details[0]['dtype'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "ngCAKLrUeEqM",
        "outputId": "3a68f681-b987-4b94-b655-365fb3d34e68"
      },
      "source": [
        "#Making Predictions on Baseline TF-Lite Model\n",
        "interpreter.set_tensor(input_details[0]['index'], images)\n",
        "interpreter.invoke()\n",
        "\n",
        "tflite_model_predictions = interpreter.get_tensor(output_details[0]['index'])\n",
        "\n",
        "print(\"Prediction results shape:\", tflite_model_predictions.shape)\n",
        "prediction_classes = np.argmax(tflite_model_predictions, axis=1)\n",
        "expected_classes = np.argmax(labels, axis=1)\n",
        "\n",
        "acc = accuracy_score(prediction_classes, expected_classes)\n",
        "print('Test accuracy TFLITE model :', acc)"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Prediction results shape: (128, 3)\n",
            "Test accuracy TFLITE model : 0.9609375\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZfiMd06geGgz"
      },
      "source": [
        "#Check the Tensors Shape - Quantise Model\n",
        "interpreter = tf.lite.Interpreter(model_path = TF_LITE_QUANT_MODEL_FILE_NAME)\n",
        "input_details = interpreter.get_input_details()\n",
        "output_details = interpreter.get_output_details()\n",
        "\n",
        "print(\"Input Shape:\", input_details[0]['shape'])\n",
        "print(\"Input Type:\", input_details[0]['dtype'])\n",
        "print(\"Output Shape:\", output_details[0]['shape'])\n",
        "print(\"Output Type:\", output_details[0]['dtype'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "Kg07MTHPeIMt",
        "outputId": "fe03ba9f-b520-4413-ecbe-1145cba02be7"
      },
      "source": [
        "#Resize the Tensor Shape - Quantise Model\n",
        "interpreter.resize_tensor_input(input_details[0]['index'], (128, 150, 150, 3))\n",
        "interpreter.resize_tensor_input(output_details[0]['index'], (128, 3))\n",
        "interpreter.allocate_tensors()\n",
        "\n",
        "input_details = interpreter.get_input_details()\n",
        "output_details = interpreter.get_output_details()\n",
        "\n",
        "print(\"Input Shape:\", input_details[0]['shape'])\n",
        "print(\"Input Type:\", input_details[0]['dtype'])\n",
        "print(\"Output Shape:\", output_details[0]['shape'])\n",
        "print(\"Output Type:\", output_details[0]['dtype'])"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Input Shape: [128 150 150   3]\n",
            "Input Type: <class 'numpy.float32'>\n",
            "Output Shape: [128   3]\n",
            "Output Type: <class 'numpy.float32'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "wNQOCFBNeJhH",
        "outputId": "799bcb7d-2f14-4008-dc38-0fa95abb958f"
      },
      "source": [
        "#Making Predictions on Quantise TF-Lite Model\n",
        "interpreter.set_tensor(input_details[0]['index'], images)\n",
        "interpreter.invoke()\n",
        "\n",
        "tflite_model_predictions = interpreter.get_tensor(output_details[0]['index'])\n",
        "\n",
        "print(\"Prediction results shape:\", tflite_model_predictions.shape)\n",
        "prediction_classes = np.argmax(tflite_model_predictions, axis=1)\n",
        "expected_classes = np.argmax(labels, axis=1)\n",
        "\n",
        "acc = accuracy_score(prediction_classes, expected_classes)\n",
        "print('Test accuracy TFLITE model :', acc)"
      ],
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Prediction results shape: (128, 3)\n",
            "Test accuracy TFLITE model : 0.9375\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "X-9Bv9I5Fa-f"
      },
      "source": [
        "# Result Summary\n",
        "\n",
        "\n",
        "| Configuration          | Baseline Model | Quantised Model |\n",
        "|------------------------|----------------|-----------------|\n",
        "| EPOCHS                 | 20             | 20              |\n",
        "| Model Accuracy         | 85.75%         | 97.84%          |\n",
        "| TF-Lite File Size      | 13.255 MB      | 3.33 MB         |\n",
        "| TF-Lite Model Accuracy | 81.25%         | 96.09%          |"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u6JEdr41eNZw"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}