Emotion_Detection(vgg_pre_trained_).ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "Emotion Detection(vgg - pre-trained )",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "eXHGVAlCGxt4",
        "outputId": "084b17a9-1fca-4362-ea10-c5e75a01c938"
      },
      "source": [
        "\n",
        "#Importing Libraries\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import tensorflow as tf\n",
        "\n",
        "from keras.layers import Conv2D,Input, ZeroPadding2D, BatchNormalization, Activation, MaxPool2D, Flatten,Dense,Dropout,GlobalAveragePooling2D\n",
        "from keras.models import Sequential  \n",
        "from keras.models import Model,load_model\n",
        "from keras.callbacks import ReduceLROnPlateau, ModelCheckpoint\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import f1_score\n",
        "\n",
        "\n",
        "import time\n",
        "import matplotlib.pyplot as plt\n",
        "import os\n",
        "from google.colab import drive\n",
        "drive.mount(\"/content/drive\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DNThlbBghcHK"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WjrhG7XZojTt"
      },
      "source": [
        "#Reading The Data\n",
        "data = pd.read_csv(\"/content/drive/My Drive/Data/fer2013.csv\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-7xmgq-r-cRN",
        "outputId": "f7e8c2c3-d0ef-4f51-d92e-fc38e1060a19"
      },
      "source": [
        "data.dtypes"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "emotion     int64\n",
              "pixels     object\n",
              "Usage      object\n",
              "dtype: object"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "y3Wg5mZzhs-h"
      },
      "source": [
        "# #Data Preprocessing\n",
        "\n",
        "# #Stacking\n",
        "# data[\"pixel\"]=data[\"pixels\"]+\" \"+data[\"pixels\"]+\" \"+data[\"pixels\"]\n",
        "# del data[\"pixels\"]\n",
        "\n",
        "\n",
        "width, height = 48, 48\n",
        "\n",
        "datapoints = data['pixels'].tolist()\n",
        "\n",
        "#getting features for training\n",
        "X = []\n",
        "for xseq in datapoints:\n",
        "    xx = [int(xp) for xp in xseq.split(' ')]\n",
        "    xx = np.asarray(xx).reshape(width, height,1)\n",
        "    X.append(xx.astype('float32'))\n",
        "\n",
        "X = np.asarray(X)\n",
        "\n",
        "#getting labels for training\n",
        "y = pd.get_dummies(data['emotion']).values"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_hXFEZ32fnBO",
        "outputId": "a8a3a9c9-b4cd-477b-d2e3-c41224bcdb66"
      },
      "source": [
        "print(f\"X shape : {X.shape}\")\n",
        "print(f\"y shape : {y.shape}\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "X shape : (35887, 48, 48, 1)\n",
            "y shape : (35887, 7)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8zyl-MQMnWL4"
      },
      "source": [
        "#Spliting the Dataset \n",
        "X_train, X_test_val, y_train, y_test_val = train_test_split(X, y, test_size=0.3,random_state=2)\n",
        "X_test, X_val, y_test, y_val = train_test_split(X_test_val, y_test_val, test_size=0.5,random_state=2)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uzkjKBp11QGO",
        "outputId": "bda8bfbe-44fc-4bca-96a8-865398353c76"
      },
      "source": [
        "print(f\"X_train shape : {X_train.shape}\")\n",
        "print(f\"y_train shape : {y_train.shape}\")\n",
        "print(f\"X_val shape : {X_val.shape}\")\n",
        "print(f\"y_val shape : {y_val.shape}\")\n",
        "print(f\"X_test shape : {X_test.shape}\")\n",
        "print(f\"y_test shape : {y_test.shape}\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "X_train shape : (25120, 48, 48, 1)\n",
            "y_train shape : (25120, 7)\n",
            "X_val shape : (5384, 48, 48, 1)\n",
            "y_val shape : (5384, 7)\n",
            "X_test shape : (5383, 48, 48, 1)\n",
            "y_test shape : (5383, 7)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LTDPCYGDmCFS",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "13cbd1ab-2c99-4f17-fd5e-9edfae38a98f"
      },
      "source": [
        "# from keras.applications.vgg16 import VGG16\n",
        "# #Transfer Learning\n",
        "# conv_base = VGG16(include_top=False,\n",
        "#                       weights='imagenet',\n",
        "#                       input_shape=(48,48, 3))\n",
        "# for layer in conv_base.layers[:-2]:\n",
        "#  layer.trainable = False\n",
        " \n",
        "# model = Sequential()\n",
        "# model.add(conv_base)\n",
        "# model.add(GlobalAveragePooling2D())\n",
        "# model.add(Dense(512, activation='relu'))\n",
        "# model.add(BatchNormalization())\n",
        "# model.add(Dropout(0.5))\n",
        "# model.add(Dense(128, activation='relu'))\n",
        "# model.add(BatchNormalization())\n",
        "# model.add(Dropout(0.5))\n",
        "# # model.add(Dense(64, activation='relu'))\n",
        "# # model.add(BatchNormalization())\n",
        "# model.add(Dense(32, activation='relu'))\n",
        "# # model.add(Dense(16, activation='relu'))\n",
        "# model.add(Dense(7, activation='softmax'))\n",
        "# # import keras\n",
        "# # EPOCHS = 50\n",
        "# # BS = 128\n",
        "# # DROPOUT_RATE = 0.5\n",
        "# # FROZEN_LAYER_NUM = 19\n",
        "\n",
        "# # ADAM_LEARNING_RATE = 0.001\n",
        "# # SGD_LEARNING_RATE = 0.01\n",
        "# # SGD_DECAY = 0.0001\n",
        "\n",
        "\n",
        "# model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])\n",
        "# vgg_notop = VGGFace(model='vgg16', include_top=False, input_shape=(48,48, 3), pooling='avg')\n",
        "# #print(vgg_notop.summary())\n",
        "# last_layer = vgg_notop.get_layer('pool5').output\n",
        "# x = Flatten(name='flatten')(last_layer)\n",
        "# x = Dropout(DROPOUT_RATE)(x)\n",
        "# x = Dense(4096, activation='relu', name='fc6')(x)\n",
        "# x = Dropout(DROPOUT_RATE)(x)\n",
        "# x = Dense(1024, activation='relu', name='fc7')(x)\n",
        "\n",
        "# for i in range(19):\n",
        "#     vgg_notop.layers[i].trainable = False\n",
        "\n",
        "# print(vgg_notop.get_layer('pool5').trainable)\n",
        "\n",
        "# out = Dense(7, activation='softmax', name='classifier')(x)\n",
        "\n",
        "# model = Model(vgg_notop.input, out)\n",
        "\n",
        "\n",
        "# # optim = keras.optimizers.Adam(lr=ADAM_LEARNING_RATE, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)\n",
        "# # #optim = keras.optimizers.Adam(lr=0.0005, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)\n",
        "# # sgd = keras.optimizers.SGD(lr=SGD_LEARNING_RATE, momentum=0.9, decay=SGD_DECAY, nesterov=True)\n",
        "# # rlrop = keras.callbacks.ReduceLROnPlateau(monitor='val_acc',mode='max',factor=0.5, patience=10, min_lr=0.00001, verbose=1)\n",
        "\n",
        "\n",
        "# from keras.layers import Conv2D,Input, ZeroPadding2D, BatchNormalization, Activation, MaxPool2D, Flatten,Dense,Dropout,GlobalAveragePooling2D,Convolution2D,MaxPooling2D\n",
        "\n",
        "\n",
        "# model = Sequential()\n",
        "# model.add(ZeroPadding2D((1,1),input_shape=(48,48,1)))\n",
        "# model.add(Convolution2D(64, 3, 3, activation='relu'))\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(64, 3, 3, activation='relu'))\n",
        "# model.add(MaxPooling2D((2,2), strides=(2,2)))\n",
        "\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(128, 3, 3, activation='relu'))\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(128, 3, 3, activation='relu'))\n",
        "# model.add(MaxPooling2D((2,2), strides=(2,2)))\n",
        "\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(256, 3, 3, activation='relu'))\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(256, 3, 3, activation='relu'))\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(256, 3, 3, activation='relu'))\n",
        "# model.add(MaxPooling2D((2,2), strides=(2,2)))\n",
        "\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(512, 3, 3, activation='relu'))\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(512, 3, 3, activation='relu'))\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(512, 3, 3, activation='relu'))\n",
        "# model.add(MaxPooling2D((2,2), strides=(2,2)))\n",
        "\n",
        "\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(512, 3, 3, activation='relu'))\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(512, 3, 3, activation='relu'))\n",
        "# model.add(ZeroPadding2D((1,1)))\n",
        "# model.add(Convolution2D(512, 3, 3, activation='relu'))\n",
        "# model.add(MaxPooling2D((2,2), strides=(2,2)))\n",
        "\n",
        "# model.add(Flatten())\n",
        "# model.add(Dense(4096, activation='relu'))\n",
        "# model.add(Dropout(0.5))\n",
        "# model.add(Dense(4096, activation='relu'))\n",
        "# model.add(Dropout(0.5))\n",
        "# model.add(Dense(7, activation='softmax'))\n",
        "# model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])\n",
        "np.random.seed(10)\n",
        "model = Sequential()\n",
        "\n",
        "# 1. Conv\n",
        "model.add(Conv2D(64, (3, 3), padding='same', input_shape=(48,48,1)))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.25))\n",
        "\n",
        "\n",
        "# 2. Conv Layer\n",
        "model.add(Conv2D(128, (5, 5), padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.25))\n",
        "\n",
        "# 3. Conv Layer\n",
        "model.add(Conv2D(512, (3,3), padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.25))\n",
        "\n",
        "#4. Conv Layer\n",
        "model.add(Conv2D(512, (3,3), padding='same'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.25))\n",
        "\n",
        "model.add(Flatten())\n",
        "model.add(Dense(256))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(Dropout(0.25))\n",
        "\n",
        "model.add(Dense(512))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Activation('relu'))\n",
        "model.add(Dropout(0.25))\n",
        "\n",
        "model.add(Dense(7, activation='softmax'))\n",
        "model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_5\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_20 (Conv2D)           (None, 48, 48, 64)        640       \n",
            "_________________________________________________________________\n",
            "batch_normalization_24 (Batc (None, 48, 48, 64)        256       \n",
            "_________________________________________________________________\n",
            "activation_24 (Activation)   (None, 48, 48, 64)        0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_18 (MaxPooling (None, 24, 24, 64)        0         \n",
            "_________________________________________________________________\n",
            "dropout_24 (Dropout)         (None, 24, 24, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_21 (Conv2D)           (None, 24, 24, 128)       204928    \n",
            "_________________________________________________________________\n",
            "batch_normalization_25 (Batc (None, 24, 24, 128)       512       \n",
            "_________________________________________________________________\n",
            "activation_25 (Activation)   (None, 24, 24, 128)       0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_19 (MaxPooling (None, 12, 12, 128)       0         \n",
            "_________________________________________________________________\n",
            "dropout_25 (Dropout)         (None, 12, 12, 128)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_22 (Conv2D)           (None, 12, 12, 512)       590336    \n",
            "_________________________________________________________________\n",
            "batch_normalization_26 (Batc (None, 12, 12, 512)       2048      \n",
            "_________________________________________________________________\n",
            "activation_26 (Activation)   (None, 12, 12, 512)       0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_20 (MaxPooling (None, 6, 6, 512)         0         \n",
            "_________________________________________________________________\n",
            "dropout_26 (Dropout)         (None, 6, 6, 512)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_23 (Conv2D)           (None, 6, 6, 512)         2359808   \n",
            "_________________________________________________________________\n",
            "batch_normalization_27 (Batc (None, 6, 6, 512)         2048      \n",
            "_________________________________________________________________\n",
            "activation_27 (Activation)   (None, 6, 6, 512)         0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_21 (MaxPooling (None, 3, 3, 512)         0         \n",
            "_________________________________________________________________\n",
            "dropout_27 (Dropout)         (None, 3, 3, 512)         0         \n",
            "_________________________________________________________________\n",
            "flatten_4 (Flatten)          (None, 4608)              0         \n",
            "_________________________________________________________________\n",
            "dense_12 (Dense)             (None, 256)               1179904   \n",
            "_________________________________________________________________\n",
            "batch_normalization_28 (Batc (None, 256)               1024      \n",
            "_________________________________________________________________\n",
            "activation_28 (Activation)   (None, 256)               0         \n",
            "_________________________________________________________________\n",
            "dropout_28 (Dropout)         (None, 256)               0         \n",
            "_________________________________________________________________\n",
            "dense_13 (Dense)             (None, 512)               131584    \n",
            "_________________________________________________________________\n",
            "batch_normalization_29 (Batc (None, 512)               2048      \n",
            "_________________________________________________________________\n",
            "activation_29 (Activation)   (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dropout_29 (Dropout)         (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_14 (Dense)             (None, 7)                 3591      \n",
            "=================================================================\n",
            "Total params: 4,478,727\n",
            "Trainable params: 4,474,759\n",
            "Non-trainable params: 3,968\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "W2JeQkplnCYQ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "84969edc-4c03-42c8-ab77-81f7f115d611"
      },
      "source": [
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_8 (Conv2D)            (None, 48, 48, 64)        640       \n",
            "_________________________________________________________________\n",
            "batch_normalization_6 (Batch (None, 48, 48, 64)        256       \n",
            "_________________________________________________________________\n",
            "activation_6 (Activation)    (None, 48, 48, 64)        0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_6 (MaxPooling2 (None, 24, 24, 64)        0         \n",
            "_________________________________________________________________\n",
            "dropout_6 (Dropout)          (None, 24, 24, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_9 (Conv2D)            (None, 24, 24, 128)       204928    \n",
            "_________________________________________________________________\n",
            "batch_normalization_7 (Batch (None, 24, 24, 128)       512       \n",
            "_________________________________________________________________\n",
            "activation_7 (Activation)    (None, 24, 24, 128)       0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_7 (MaxPooling2 (None, 12, 12, 128)       0         \n",
            "_________________________________________________________________\n",
            "dropout_7 (Dropout)          (None, 12, 12, 128)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_10 (Conv2D)           (None, 12, 12, 512)       590336    \n",
            "_________________________________________________________________\n",
            "batch_normalization_8 (Batch (None, 12, 12, 512)       2048      \n",
            "_________________________________________________________________\n",
            "activation_8 (Activation)    (None, 12, 12, 512)       0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_8 (MaxPooling2 (None, 6, 6, 512)         0         \n",
            "_________________________________________________________________\n",
            "dropout_8 (Dropout)          (None, 6, 6, 512)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_11 (Conv2D)           (None, 6, 6, 512)         2359808   \n",
            "_________________________________________________________________\n",
            "batch_normalization_9 (Batch (None, 6, 6, 512)         2048      \n",
            "_________________________________________________________________\n",
            "activation_9 (Activation)    (None, 6, 6, 512)         0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_9 (MaxPooling2 (None, 3, 3, 512)         0         \n",
            "_________________________________________________________________\n",
            "dropout_9 (Dropout)          (None, 3, 3, 512)         0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 4608)              0         \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 256)               1179904   \n",
            "_________________________________________________________________\n",
            "batch_normalization_10 (Batc (None, 256)               1024      \n",
            "_________________________________________________________________\n",
            "activation_10 (Activation)   (None, 256)               0         \n",
            "_________________________________________________________________\n",
            "dropout_10 (Dropout)         (None, 256)               0         \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 512)               131584    \n",
            "_________________________________________________________________\n",
            "batch_normalization_11 (Batc (None, 512)               2048      \n",
            "_________________________________________________________________\n",
            "activation_11 (Activation)   (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dropout_11 (Dropout)         (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 7)                 3591      \n",
            "=================================================================\n",
            "Total params: 4,478,727\n",
            "Trainable params: 4,474,759\n",
            "Non-trainable params: 3,968\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fykfZ5ZWC6ec"
      },
      "source": [
        "#CallBacks\n",
        "\n",
        "# reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2,\n",
        "#                               patience=5, min_lr=0.001)\n",
        "\n",
        "\n",
        "\n",
        "# checkpoint_filepath = '/content/drive/My Drive/Data/checkpoints/weights-improvement-{epoch:02d}-{val_accuracy:.2f}.h5'\n",
        "\n",
        "# checkpoints = ModelCheckpoint(filepath=checkpoint_filepath,\n",
        "#                              monitor=\"val_accuracy\",\n",
        "#                              verbose=1,\n",
        "#                              save_best_only=True,\n",
        "#                              mode='auto',\n",
        "#                              save_freq=\"epoch\"\n",
        "# #                              )\n",
        "# checkpoint=[reduce_lr]\n",
        "#,callbacks=[checkpoint]\n",
        "# from keras.preprocessing.image import ImageDataGenerator\n",
        "# datagen = ImageDataGenerator(\n",
        "#     featurewise_center=True,\n",
        "#     featurewise_std_normalization=True,\n",
        "#     rotation_range=20,\n",
        "#     width_shift_range=0.2,\n",
        "#     height_shift_range=0.2,\n",
        "#     horizontal_flip=True)\n",
        "# datagen.fit(X_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oL979KR5n7f8",
        "outputId": "688d1956-6ee7-4220-8ffb-23ad7d8b64ec"
      },
      "source": [
        "history=model.fit(x=X_train, y=y_train, batch_size=128, epochs=10, validation_data=(X_val, y_val))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "197/197 [==============================] - 11s 52ms/step - loss: 1.9459 - accuracy: 0.2585 - val_loss: 1.8673 - val_accuracy: 0.2964\n",
            "Epoch 2/10\n",
            "197/197 [==============================] - 10s 51ms/step - loss: 1.5216 - accuracy: 0.4171 - val_loss: 1.4826 - val_accuracy: 0.4307\n",
            "Epoch 3/10\n",
            "197/197 [==============================] - 10s 52ms/step - loss: 1.3300 - accuracy: 0.4886 - val_loss: 1.3800 - val_accuracy: 0.4785\n",
            "Epoch 4/10\n",
            "197/197 [==============================] - 10s 51ms/step - loss: 1.2146 - accuracy: 0.5375 - val_loss: 1.4663 - val_accuracy: 0.3936\n",
            "Epoch 5/10\n",
            "197/197 [==============================] - 10s 51ms/step - loss: 1.1355 - accuracy: 0.5690 - val_loss: 1.2696 - val_accuracy: 0.5249\n",
            "Epoch 6/10\n",
            "197/197 [==============================] - 10s 50ms/step - loss: 1.0876 - accuracy: 0.5928 - val_loss: 1.2601 - val_accuracy: 0.5199\n",
            "Epoch 7/10\n",
            "197/197 [==============================] - 10s 50ms/step - loss: 1.0337 - accuracy: 0.6097 - val_loss: 1.1986 - val_accuracy: 0.5579\n",
            "Epoch 8/10\n",
            "197/197 [==============================] - 10s 50ms/step - loss: 0.9909 - accuracy: 0.6255 - val_loss: 1.1184 - val_accuracy: 0.5775\n",
            "Epoch 9/10\n",
            "197/197 [==============================] - 10s 50ms/step - loss: 0.9450 - accuracy: 0.6435 - val_loss: 1.3309 - val_accuracy: 0.4900\n",
            "Epoch 10/10\n",
            "197/197 [==============================] - 10s 51ms/step - loss: 0.8977 - accuracy: 0.6663 - val_loss: 1.0698 - val_accuracy: 0.5925\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z0LSXWOZrvcs"
      },
      "source": [
        " def plot_metrics(history):\n",
        "    \n",
        "    train_loss = history['loss']\n",
        "    val_loss = history['val_loss']\n",
        "    train_acc = history['accuracy']\n",
        "    val_acc = history['val_accuracy']\n",
        "    \n",
        "    # Loss\n",
        "    plt.figure()\n",
        "    plt.plot(train_loss, label='Training Loss')\n",
        "    plt.plot(val_loss, label='Validation Loss')\n",
        "    plt.title('Loss')\n",
        "    plt.legend()\n",
        "    plt.show()\n",
        "    \n",
        "    # Accuracy\n",
        "    plt.figure()\n",
        "    plt.plot(train_acc, label='Training Accuracy')\n",
        "    plt.plot(val_acc, label='Validation Accuracy')\n",
        "    plt.title('Accuracy')\n",
        "    plt.legend()\n",
        "    plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "My68sr14swGO",
        "outputId": "ae5ed105-8c7b-4f0c-8179-d13752dd0fcf"
      },
      "source": [
        "history = model.history.history\n",
        "plot_metrics(history)"
      ],
      "execution_count": null,
      "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": "code",
      "metadata": {
        "id": "zQF71t_v7Pir"
      },
      "source": [
        "#ISSUES:: Overfitting :use dropout and increase validation dataset and shuffle before train_test_split \n",
        "\n",
        "# bestVGG16_model = load_model(filepath='/content/gdrive/My Drive/VGGmodels/075-0.99')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Fw75tFNi7aZB"
      },
      "source": [
        "# loss, acc = bestVGG16_model.evaluate(x=X_test, y=Y_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OUsGUSbm9Ihz"
      },
      "source": [
        "# reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2,\n",
        "#                               patience=5, min_lr=0.001)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4GUCCbUIaiQC"
      },
      "source": [
        "# from keras.applications import ResNet50\n",
        "# INCV3=ResNet50(include_top=False, weights='imagenet',  input_shape=(48,48,3),pooling=max)\n",
        "# incv3=Sequential()\n",
        "# incv3.add(INCV3)\n",
        "# incv3.add(Dropout(0.3))\n",
        "# incv3.add(Flatten())\n",
        "# incv3.add(Dropout(0.5))\n",
        "# incv3.add(Dense(7, activation='relu'))\n",
        "# incv3.layers[0].trainable = False\n",
        "# incv3.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dmpmbwpPbBko",
        "outputId": "7716dcd4-1b1b-45b3-85f6-e79b22263c9e"
      },
      "source": [
        "# incv3.fit(X_train, y_train, batch_size=128,validation_data=(X_val, y_val), epochs = 40, verbose = 1,shuffle=True)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3650 - accuracy: 0.3130 - val_loss: 3.3415 - val_accuracy: 0.2933\n",
            "Epoch 2/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3627 - accuracy: 0.3136 - val_loss: 3.3316 - val_accuracy: 0.2949\n",
            "Epoch 3/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3672 - accuracy: 0.3109 - val_loss: 3.3479 - val_accuracy: 0.3013\n",
            "Epoch 4/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3645 - accuracy: 0.3170 - val_loss: 3.3326 - val_accuracy: 0.2962\n",
            "Epoch 5/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3587 - accuracy: 0.3162 - val_loss: 3.3447 - val_accuracy: 0.3022\n",
            "Epoch 6/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3568 - accuracy: 0.3194 - val_loss: 3.3430 - val_accuracy: 0.2966\n",
            "Epoch 7/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3618 - accuracy: 0.3139 - val_loss: 3.3327 - val_accuracy: 0.2970\n",
            "Epoch 8/40\n",
            "197/197 [==============================] - 17s 85ms/step - loss: 3.3606 - accuracy: 0.3174 - val_loss: 3.3377 - val_accuracy: 0.2981\n",
            "Epoch 9/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3584 - accuracy: 0.3156 - val_loss: 3.3390 - val_accuracy: 0.2951\n",
            "Epoch 10/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3611 - accuracy: 0.3170 - val_loss: 3.3482 - val_accuracy: 0.3024\n",
            "Epoch 11/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3596 - accuracy: 0.3169 - val_loss: 3.3407 - val_accuracy: 0.3016\n",
            "Epoch 12/40\n",
            "197/197 [==============================] - 16s 83ms/step - loss: 3.3583 - accuracy: 0.3188 - val_loss: 3.3421 - val_accuracy: 0.2962\n",
            "Epoch 13/40\n",
            "197/197 [==============================] - 16s 84ms/step - loss: 3.3558 - accuracy: 0.3172 - val_loss: 3.3410 - val_accuracy: 0.2981\n",
            "Epoch 14/40\n",
            "197/197 [==============================] - 16s 84ms/step - loss: 3.3499 - accuracy: 0.3199 - val_loss: 3.3547 - val_accuracy: 0.3024\n",
            "Epoch 15/40\n",
            "197/197 [==============================] - 16s 83ms/step - loss: 3.3594 - accuracy: 0.3206 - val_loss: 3.3552 - val_accuracy: 0.3001\n",
            "Epoch 16/40\n",
            "197/197 [==============================] - 16s 83ms/step - loss: 3.3520 - accuracy: 0.3174 - val_loss: 3.3515 - val_accuracy: 0.2942\n",
            "Epoch 17/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3545 - accuracy: 0.3201 - val_loss: 3.3650 - val_accuracy: 0.3024\n",
            "Epoch 18/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3524 - accuracy: 0.3184 - val_loss: 3.3493 - val_accuracy: 0.2964\n",
            "Epoch 19/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3529 - accuracy: 0.3206 - val_loss: 3.3457 - val_accuracy: 0.2998\n",
            "Epoch 20/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3565 - accuracy: 0.3195 - val_loss: 3.3588 - val_accuracy: 0.2990\n",
            "Epoch 21/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3495 - accuracy: 0.3227 - val_loss: 3.3524 - val_accuracy: 0.3031\n",
            "Epoch 22/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3481 - accuracy: 0.3225 - val_loss: 3.3513 - val_accuracy: 0.2990\n",
            "Epoch 23/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3522 - accuracy: 0.3181 - val_loss: 3.3349 - val_accuracy: 0.2979\n",
            "Epoch 24/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3516 - accuracy: 0.3204 - val_loss: 3.3386 - val_accuracy: 0.2883\n",
            "Epoch 25/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3472 - accuracy: 0.3207 - val_loss: 3.3368 - val_accuracy: 0.2979\n",
            "Epoch 26/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3522 - accuracy: 0.3215 - val_loss: 3.3305 - val_accuracy: 0.2909\n",
            "Epoch 27/40\n",
            "197/197 [==============================] - 17s 85ms/step - loss: 3.3509 - accuracy: 0.3185 - val_loss: 3.3504 - val_accuracy: 0.2985\n",
            "Epoch 28/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3453 - accuracy: 0.3202 - val_loss: 3.3655 - val_accuracy: 0.3018\n",
            "Epoch 29/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3472 - accuracy: 0.3215 - val_loss: 3.3507 - val_accuracy: 0.3011\n",
            "Epoch 30/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3545 - accuracy: 0.3211 - val_loss: 3.3828 - val_accuracy: 0.2987\n",
            "Epoch 31/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3496 - accuracy: 0.3234 - val_loss: 3.3679 - val_accuracy: 0.2955\n",
            "Epoch 32/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3526 - accuracy: 0.3264 - val_loss: 3.3590 - val_accuracy: 0.3013\n",
            "Epoch 33/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3468 - accuracy: 0.3208 - val_loss: 3.3616 - val_accuracy: 0.3046\n",
            "Epoch 34/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3480 - accuracy: 0.3246 - val_loss: 3.3548 - val_accuracy: 0.3024\n",
            "Epoch 35/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3518 - accuracy: 0.3232 - val_loss: 3.3402 - val_accuracy: 0.3016\n",
            "Epoch 36/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3431 - accuracy: 0.3240 - val_loss: 3.3367 - val_accuracy: 0.2981\n",
            "Epoch 37/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3542 - accuracy: 0.3221 - val_loss: 3.3640 - val_accuracy: 0.2996\n",
            "Epoch 38/40\n",
            "197/197 [==============================] - 17s 84ms/step - loss: 3.3438 - accuracy: 0.3277 - val_loss: 3.3561 - val_accuracy: 0.3018\n",
            "Epoch 39/40\n",
            "197/197 [==============================] - 16s 84ms/step - loss: 3.3464 - accuracy: 0.3270 - val_loss: 3.3492 - val_accuracy: 0.3029\n",
            "Epoch 40/40\n",
            "197/197 [==============================] - 16s 83ms/step - loss: 3.3492 - accuracy: 0.3242 - val_loss: 3.3524 - val_accuracy: 0.2990\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f79fb476828>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2ycXV-ZJclTE"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}