facial_emotion_recognition.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "facial emotion recognition.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "0JyAVtXyB_yN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import os\n",
        "import cv2"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nX9OJo78aUsh",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "1562dae3-f9cf-425b-a5f7-fcfcaa262e68"
      },
      "source": [
        "import tensorflow as tf\n",
        "tf.__version__"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'2.2.0-rc3'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 2
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jlc3AjZ1TWTJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from tensorflow.keras.layers import Dense,Dropout,Flatten\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "from tensorflow.keras.layers import Convolution2D\n",
        "from tensorflow.keras.layers import MaxPooling2D"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vmLMUyvnsxwV",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.metrics import classification_report"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cLcTEE0TbTJj",
        "colab_type": "text"
      },
      "source": [
        "## Building a CNN Model\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ljvFcTL1auiG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model=Sequential()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "O35A4Q_2bcPO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.add(Convolution2D(64,(5,5),input_shape=(48,48,1),activation='relu'))\n",
        "model.add(Convolution2D(64,(5,5),activation='relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.4))\n",
        "model.add(Convolution2D(128,(3,3),activation='relu'))\n",
        "model.add(Convolution2D(128,(3,3),activation='relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\n",
        "model.add(Dropout(0.4))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(units=128,activation='relu'))\n",
        "model.add(Dense(units=7,activation='softmax'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kw3pfdqkfh7y",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 542
        },
        "outputId": "9aedbc77-0100-428a-f37b-6922992b1977"
      },
      "source": [
        "model.summary()"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 44, 44, 64)        1664      \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 40, 40, 64)        102464    \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 20, 20, 64)        0         \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 20, 20, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 18, 18, 128)       73856     \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 16, 16, 128)       147584    \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 8, 8, 128)         0         \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 8, 8, 128)         0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 8192)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 128)               1048704   \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 7)                 903       \n",
            "=================================================================\n",
            "Total params: 1,375,175\n",
            "Trainable params: 1,375,175\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kzybn_q9kFBz",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pga7bfXCC_OJ",
        "colab_type": "text"
      },
      "source": [
        "## Image Argumentation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "K7Ok79-_g0xy",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "outputId": "d27d2c13-0044-4394-8137-0907ee86bbee"
      },
      "source": [
        "train_datagen = ImageDataGenerator(\n",
        "        rescale=1./255,\n",
        "        rotation_range=30,\n",
        "        shear_range=0.3,\n",
        "        zoom_range=0.3,\n",
        "        horizontal_flip=False)\n",
        "\n",
        "test_datagen = ImageDataGenerator(rescale=1./255)\n",
        "\n",
        "train_generator = train_datagen.flow_from_directory(\n",
        "        '/content/Training',\n",
        "        target_size=(48,48),\n",
        "        batch_size=512,\n",
        "        color_mode=\"grayscale\",\n",
        "        class_mode='categorical')\n",
        "\n",
        "validation_generator = test_datagen.flow_from_directory(\n",
        "        '/content/Testing',\n",
        "        target_size=(48,48),\n",
        "        batch_size=512,\n",
        "        color_mode=\"grayscale\",\n",
        "        class_mode='categorical')\n"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 28709 images belonging to 7 classes.\n",
            "Found 7178 images belonging to 7 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IC3QundqkxHO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "filepath = os.path.join(\"./emotion_detector_models/model_v{epoch}.hdf5\")\n",
        "\n",
        "checkpoint = tf.keras.callbacks.ModelCheckpoint(filepath,\n",
        "                                             monitor='val_accuracy',\n",
        "                                             verbose=1,\n",
        "                                             save_best_only=True,\n",
        "                                             mode='max')\n",
        "callbacks = [checkpoint]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4vUL74jxl8xg",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "nb_train_samples = 28709\n",
        "nb_validation_samples = 7178\n",
        "batch_size=512"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e4v11eUkj3jS",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "cd3db206-49ec-4a6a-bac2-5228b631ab56"
      },
      "source": [
        "history=model.fit_generator(\n",
        "        train_generator,\n",
        "        steps_per_epoch=nb_train_samples// batch_size,\n",
        "        epochs=50,\n",
        "        validation_data=validation_generator,\n",
        "        callbacks = callbacks,\n",
        "        validation_steps=nb_validation_samples // batch_size)"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From <ipython-input-12-059c04d1960f>:7: Model.fit_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use Model.fit, which supports generators.\n",
            "Epoch 1/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.8276 - accuracy: 0.2423\n",
            "Epoch 00001: val_accuracy improved from -inf to 0.24777, saving model to ./emotion_detector_models/model_v_1.hdf5\n",
            "56/56 [==============================] - 18s 314ms/step - loss: 1.8276 - accuracy: 0.2423 - val_loss: 1.8138 - val_accuracy: 0.2478\n",
            "Epoch 2/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.7829 - accuracy: 0.2603\n",
            "Epoch 00002: val_accuracy improved from 0.24777 to 0.29548, saving model to ./emotion_detector_models/model_v_2.hdf5\n",
            "56/56 [==============================] - 18s 322ms/step - loss: 1.7829 - accuracy: 0.2603 - val_loss: 1.7338 - val_accuracy: 0.2955\n",
            "Epoch 3/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.7390 - accuracy: 0.2873\n",
            "Epoch 00003: val_accuracy improved from 0.29548 to 0.30539, saving model to ./emotion_detector_models/model_v_3.hdf5\n",
            "56/56 [==============================] - 18s 320ms/step - loss: 1.7390 - accuracy: 0.2873 - val_loss: 1.7157 - val_accuracy: 0.3054\n",
            "Epoch 4/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.7190 - accuracy: 0.2997\n",
            "Epoch 00004: val_accuracy improved from 0.30539 to 0.34068, saving model to ./emotion_detector_models/model_v_4.hdf5\n",
            "56/56 [==============================] - 18s 313ms/step - loss: 1.7190 - accuracy: 0.2997 - val_loss: 1.6582 - val_accuracy: 0.3407\n",
            "Epoch 5/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.6993 - accuracy: 0.3114\n",
            "Epoch 00005: val_accuracy improved from 0.34068 to 0.36761, saving model to ./emotion_detector_models/model_v_5.hdf5\n",
            "56/56 [==============================] - 18s 313ms/step - loss: 1.6993 - accuracy: 0.3114 - val_loss: 1.6181 - val_accuracy: 0.3676\n",
            "Epoch 6/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.6599 - accuracy: 0.3429\n",
            "Epoch 00006: val_accuracy improved from 0.36761 to 0.40960, saving model to ./emotion_detector_models/model_v_6.hdf5\n",
            "56/56 [==============================] - 18s 314ms/step - loss: 1.6599 - accuracy: 0.3429 - val_loss: 1.5518 - val_accuracy: 0.4096\n",
            "Epoch 7/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.6259 - accuracy: 0.3621\n",
            "Epoch 00007: val_accuracy improved from 0.40960 to 0.42020, saving model to ./emotion_detector_models/model_v_7.hdf5\n",
            "56/56 [==============================] - 18s 313ms/step - loss: 1.6259 - accuracy: 0.3621 - val_loss: 1.5101 - val_accuracy: 0.4202\n",
            "Epoch 8/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.5828 - accuracy: 0.3804\n",
            "Epoch 00008: val_accuracy improved from 0.42020 to 0.44057, saving model to ./emotion_detector_models/model_v_8.hdf5\n",
            "56/56 [==============================] - 17s 311ms/step - loss: 1.5828 - accuracy: 0.3804 - val_loss: 1.4458 - val_accuracy: 0.4406\n",
            "Epoch 9/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.5618 - accuracy: 0.3881\n",
            "Epoch 00009: val_accuracy improved from 0.44057 to 0.45480, saving model to ./emotion_detector_models/model_v_9.hdf5\n",
            "56/56 [==============================] - 17s 311ms/step - loss: 1.5618 - accuracy: 0.3881 - val_loss: 1.4139 - val_accuracy: 0.4548\n",
            "Epoch 10/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.5336 - accuracy: 0.4060\n",
            "Epoch 00010: val_accuracy improved from 0.45480 to 0.47154, saving model to ./emotion_detector_models/model_v_10.hdf5\n",
            "56/56 [==============================] - 17s 312ms/step - loss: 1.5336 - accuracy: 0.4060 - val_loss: 1.3911 - val_accuracy: 0.4715\n",
            "Epoch 11/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.4952 - accuracy: 0.4234\n",
            "Epoch 00011: val_accuracy improved from 0.47154 to 0.48744, saving model to ./emotion_detector_models/model_v_11.hdf5\n",
            "56/56 [==============================] - 17s 310ms/step - loss: 1.4952 - accuracy: 0.4234 - val_loss: 1.3454 - val_accuracy: 0.4874\n",
            "Epoch 12/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.4662 - accuracy: 0.4339\n",
            "Epoch 00012: val_accuracy improved from 0.48744 to 0.48828, saving model to ./emotion_detector_models/model_v_12.hdf5\n",
            "56/56 [==============================] - 17s 310ms/step - loss: 1.4662 - accuracy: 0.4339 - val_loss: 1.3313 - val_accuracy: 0.4883\n",
            "Epoch 13/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.4471 - accuracy: 0.4412\n",
            "Epoch 00013: val_accuracy improved from 0.48828 to 0.50084, saving model to ./emotion_detector_models/model_v_13.hdf5\n",
            "56/56 [==============================] - 17s 310ms/step - loss: 1.4471 - accuracy: 0.4412 - val_loss: 1.3010 - val_accuracy: 0.5008\n",
            "Epoch 14/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.4145 - accuracy: 0.4538\n",
            "Epoch 00014: val_accuracy improved from 0.50084 to 0.50698, saving model to ./emotion_detector_models/model_v_14.hdf5\n",
            "56/56 [==============================] - 17s 309ms/step - loss: 1.4145 - accuracy: 0.4538 - val_loss: 1.2763 - val_accuracy: 0.5070\n",
            "Epoch 15/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.3869 - accuracy: 0.4691\n",
            "Epoch 00015: val_accuracy improved from 0.50698 to 0.51367, saving model to ./emotion_detector_models/model_v_15.hdf5\n",
            "56/56 [==============================] - 17s 307ms/step - loss: 1.3869 - accuracy: 0.4691 - val_loss: 1.2569 - val_accuracy: 0.5137\n",
            "Epoch 16/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.3785 - accuracy: 0.4693\n",
            "Epoch 00016: val_accuracy improved from 0.51367 to 0.52274, saving model to ./emotion_detector_models/model_v_16.hdf5\n",
            "56/56 [==============================] - 17s 311ms/step - loss: 1.3785 - accuracy: 0.4693 - val_loss: 1.2391 - val_accuracy: 0.5227\n",
            "Epoch 17/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.3739 - accuracy: 0.4741\n",
            "Epoch 00017: val_accuracy improved from 0.52274 to 0.52651, saving model to ./emotion_detector_models/model_v_17.hdf5\n",
            "56/56 [==============================] - 17s 309ms/step - loss: 1.3739 - accuracy: 0.4741 - val_loss: 1.2314 - val_accuracy: 0.5265\n",
            "Epoch 18/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.3576 - accuracy: 0.4805\n",
            "Epoch 00018: val_accuracy improved from 0.52651 to 0.53390, saving model to ./emotion_detector_models/model_v_18.hdf5\n",
            "56/56 [==============================] - 17s 306ms/step - loss: 1.3576 - accuracy: 0.4805 - val_loss: 1.2174 - val_accuracy: 0.5339\n",
            "Epoch 19/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.3358 - accuracy: 0.4893\n",
            "Epoch 00019: val_accuracy improved from 0.53390 to 0.53725, saving model to ./emotion_detector_models/model_v_19.hdf5\n",
            "56/56 [==============================] - 17s 306ms/step - loss: 1.3358 - accuracy: 0.4893 - val_loss: 1.2120 - val_accuracy: 0.5372\n",
            "Epoch 20/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.3212 - accuracy: 0.4968\n",
            "Epoch 00020: val_accuracy improved from 0.53725 to 0.54060, saving model to ./emotion_detector_models/model_v_20.hdf5\n",
            "56/56 [==============================] - 17s 307ms/step - loss: 1.3212 - accuracy: 0.4968 - val_loss: 1.1946 - val_accuracy: 0.5406\n",
            "Epoch 21/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.3248 - accuracy: 0.4952\n",
            "Epoch 00021: val_accuracy improved from 0.54060 to 0.54185, saving model to ./emotion_detector_models/model_v_21.hdf5\n",
            "56/56 [==============================] - 17s 309ms/step - loss: 1.3248 - accuracy: 0.4952 - val_loss: 1.1847 - val_accuracy: 0.5419\n",
            "Epoch 22/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.3000 - accuracy: 0.5021\n",
            "Epoch 00022: val_accuracy improved from 0.54185 to 0.54785, saving model to ./emotion_detector_models/model_v_22.hdf5\n",
            "56/56 [==============================] - 17s 305ms/step - loss: 1.3000 - accuracy: 0.5021 - val_loss: 1.1736 - val_accuracy: 0.5479\n",
            "Epoch 23/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2880 - accuracy: 0.5092\n",
            "Epoch 00023: val_accuracy did not improve from 0.54785\n",
            "56/56 [==============================] - 17s 303ms/step - loss: 1.2880 - accuracy: 0.5092 - val_loss: 1.1845 - val_accuracy: 0.5467\n",
            "Epoch 24/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2783 - accuracy: 0.5087\n",
            "Epoch 00024: val_accuracy improved from 0.54785 to 0.55343, saving model to ./emotion_detector_models/model_v_24.hdf5\n",
            "56/56 [==============================] - 17s 308ms/step - loss: 1.2783 - accuracy: 0.5087 - val_loss: 1.1653 - val_accuracy: 0.5534\n",
            "Epoch 25/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2748 - accuracy: 0.5144\n",
            "Epoch 00025: val_accuracy improved from 0.55343 to 0.55566, saving model to ./emotion_detector_models/model_v_25.hdf5\n",
            "56/56 [==============================] - 17s 306ms/step - loss: 1.2748 - accuracy: 0.5144 - val_loss: 1.1540 - val_accuracy: 0.5557\n",
            "Epoch 26/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2563 - accuracy: 0.5224\n",
            "Epoch 00026: val_accuracy did not improve from 0.55566\n",
            "56/56 [==============================] - 17s 309ms/step - loss: 1.2563 - accuracy: 0.5224 - val_loss: 1.1534 - val_accuracy: 0.5529\n",
            "Epoch 27/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2529 - accuracy: 0.5200\n",
            "Epoch 00027: val_accuracy did not improve from 0.55566\n",
            "56/56 [==============================] - 17s 306ms/step - loss: 1.2529 - accuracy: 0.5200 - val_loss: 1.1531 - val_accuracy: 0.5551\n",
            "Epoch 28/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2464 - accuracy: 0.5256\n",
            "Epoch 00028: val_accuracy improved from 0.55566 to 0.56431, saving model to ./emotion_detector_models/model_v_28.hdf5\n",
            "56/56 [==============================] - 17s 307ms/step - loss: 1.2464 - accuracy: 0.5256 - val_loss: 1.1449 - val_accuracy: 0.5643\n",
            "Epoch 29/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2356 - accuracy: 0.5300\n",
            "Epoch 00029: val_accuracy did not improve from 0.56431\n",
            "56/56 [==============================] - 17s 307ms/step - loss: 1.2356 - accuracy: 0.5300 - val_loss: 1.1618 - val_accuracy: 0.5583\n",
            "Epoch 30/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2333 - accuracy: 0.5290\n",
            "Epoch 00030: val_accuracy improved from 0.56431 to 0.56515, saving model to ./emotion_detector_models/model_v_30.hdf5\n",
            "56/56 [==============================] - 17s 304ms/step - loss: 1.2333 - accuracy: 0.5290 - val_loss: 1.1377 - val_accuracy: 0.5652\n",
            "Epoch 31/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2296 - accuracy: 0.5323\n",
            "Epoch 00031: val_accuracy did not improve from 0.56515\n",
            "56/56 [==============================] - 17s 309ms/step - loss: 1.2296 - accuracy: 0.5323 - val_loss: 1.1305 - val_accuracy: 0.5639\n",
            "Epoch 32/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2168 - accuracy: 0.5406\n",
            "Epoch 00032: val_accuracy did not improve from 0.56515\n",
            "56/56 [==============================] - 17s 303ms/step - loss: 1.2168 - accuracy: 0.5406 - val_loss: 1.1271 - val_accuracy: 0.5640\n",
            "Epoch 33/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2115 - accuracy: 0.5403\n",
            "Epoch 00033: val_accuracy improved from 0.56515 to 0.57506, saving model to ./emotion_detector_models/model_v_33.hdf5\n",
            "56/56 [==============================] - 18s 314ms/step - loss: 1.2115 - accuracy: 0.5403 - val_loss: 1.1103 - val_accuracy: 0.5751\n",
            "Epoch 34/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2085 - accuracy: 0.5374\n",
            "Epoch 00034: val_accuracy did not improve from 0.57506\n",
            "56/56 [==============================] - 17s 307ms/step - loss: 1.2085 - accuracy: 0.5374 - val_loss: 1.1155 - val_accuracy: 0.5727\n",
            "Epoch 35/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1968 - accuracy: 0.5435\n",
            "Epoch 00035: val_accuracy improved from 0.57506 to 0.57812, saving model to ./emotion_detector_models/model_v_35.hdf5\n",
            "56/56 [==============================] - 17s 309ms/step - loss: 1.1968 - accuracy: 0.5435 - val_loss: 1.1073 - val_accuracy: 0.5781\n",
            "Epoch 36/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.2034 - accuracy: 0.5428\n",
            "Epoch 00036: val_accuracy improved from 0.57812 to 0.58161, saving model to ./emotion_detector_models/model_v_36.hdf5\n",
            "56/56 [==============================] - 17s 307ms/step - loss: 1.2034 - accuracy: 0.5428 - val_loss: 1.1114 - val_accuracy: 0.5816\n",
            "Epoch 37/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1884 - accuracy: 0.5491\n",
            "Epoch 00037: val_accuracy did not improve from 0.58161\n",
            "56/56 [==============================] - 17s 302ms/step - loss: 1.1884 - accuracy: 0.5491 - val_loss: 1.1078 - val_accuracy: 0.5755\n",
            "Epoch 38/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1851 - accuracy: 0.5493\n",
            "Epoch 00038: val_accuracy did not improve from 0.58161\n",
            "56/56 [==============================] - 17s 311ms/step - loss: 1.1851 - accuracy: 0.5493 - val_loss: 1.0984 - val_accuracy: 0.5811\n",
            "Epoch 39/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1798 - accuracy: 0.5547\n",
            "Epoch 00039: val_accuracy did not improve from 0.58161\n",
            "56/56 [==============================] - 18s 316ms/step - loss: 1.1798 - accuracy: 0.5547 - val_loss: 1.0987 - val_accuracy: 0.5794\n",
            "Epoch 40/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1798 - accuracy: 0.5520\n",
            "Epoch 00040: val_accuracy did not improve from 0.58161\n",
            "56/56 [==============================] - 17s 304ms/step - loss: 1.1798 - accuracy: 0.5520 - val_loss: 1.0991 - val_accuracy: 0.5811\n",
            "Epoch 41/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1703 - accuracy: 0.5526\n",
            "Epoch 00041: val_accuracy improved from 0.58161 to 0.58454, saving model to ./emotion_detector_models/model_v_41.hdf5\n",
            "56/56 [==============================] - 17s 303ms/step - loss: 1.1703 - accuracy: 0.5526 - val_loss: 1.0920 - val_accuracy: 0.5845\n",
            "Epoch 42/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1614 - accuracy: 0.5578\n",
            "Epoch 00042: val_accuracy did not improve from 0.58454\n",
            "56/56 [==============================] - 17s 302ms/step - loss: 1.1614 - accuracy: 0.5578 - val_loss: 1.0900 - val_accuracy: 0.5834\n",
            "Epoch 43/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1567 - accuracy: 0.5633\n",
            "Epoch 00043: val_accuracy improved from 0.58454 to 0.58859, saving model to ./emotion_detector_models/model_v_43.hdf5\n",
            "56/56 [==============================] - 17s 303ms/step - loss: 1.1567 - accuracy: 0.5633 - val_loss: 1.0906 - val_accuracy: 0.5886\n",
            "Epoch 44/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1617 - accuracy: 0.5587\n",
            "Epoch 00044: val_accuracy did not improve from 0.58859\n",
            "56/56 [==============================] - 17s 303ms/step - loss: 1.1617 - accuracy: 0.5587 - val_loss: 1.0930 - val_accuracy: 0.5840\n",
            "Epoch 45/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1549 - accuracy: 0.5600\n",
            "Epoch 00045: val_accuracy did not improve from 0.58859\n",
            "56/56 [==============================] - 17s 304ms/step - loss: 1.1549 - accuracy: 0.5600 - val_loss: 1.0845 - val_accuracy: 0.5869\n",
            "Epoch 46/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1555 - accuracy: 0.5648\n",
            "Epoch 00046: val_accuracy did not improve from 0.58859\n",
            "56/56 [==============================] - 17s 301ms/step - loss: 1.1555 - accuracy: 0.5648 - val_loss: 1.0898 - val_accuracy: 0.5848\n",
            "Epoch 47/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1501 - accuracy: 0.5674\n",
            "Epoch 00047: val_accuracy improved from 0.58859 to 0.59026, saving model to ./emotion_detector_models/model_v_47.hdf5\n",
            "56/56 [==============================] - 17s 303ms/step - loss: 1.1501 - accuracy: 0.5674 - val_loss: 1.0844 - val_accuracy: 0.5903\n",
            "Epoch 48/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1421 - accuracy: 0.5646\n",
            "Epoch 00048: val_accuracy did not improve from 0.59026\n",
            "56/56 [==============================] - 17s 304ms/step - loss: 1.1421 - accuracy: 0.5646 - val_loss: 1.0961 - val_accuracy: 0.5826\n",
            "Epoch 49/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1499 - accuracy: 0.5644\n",
            "Epoch 00049: val_accuracy did not improve from 0.59026\n",
            "56/56 [==============================] - 17s 305ms/step - loss: 1.1499 - accuracy: 0.5644 - val_loss: 1.0932 - val_accuracy: 0.5823\n",
            "Epoch 50/50\n",
            "56/56 [==============================] - ETA: 0s - loss: 1.1409 - accuracy: 0.5687\n",
            "Epoch 00050: val_accuracy did not improve from 0.59026\n",
            "56/56 [==============================] - 17s 301ms/step - loss: 1.1409 - accuracy: 0.5687 - val_loss: 1.0786 - val_accuracy: 0.5862\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ayUv9KJFoUz4",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 374
        },
        "outputId": "362183cf-59fd-422a-ddb7-eb15a3fb5c0d"
      },
      "source": [
        "fig,(ax1,ax2)=plt.subplots(nrows=1,ncols=2,figsize=(20,6))\n",
        "ax1.plot(history.history['accuracy'],label='train_accuracy')\n",
        "ax1.plot(history.history['val_accuracy'],label='test_accuracy')\n",
        "ax1.legend()\n",
        "ax2.plot(history.history['loss'],label='train_loss')\n",
        "ax2.plot(history.history['val_loss'],label='test_loss')\n",
        "ax2.legend()\n",
        "plt.show()"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x432 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tSumRgD0DE5F",
        "colab_type": "text"
      },
      "source": [
        "## Predicting And Checking Accuracy"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZI4rKmnvqypP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 139
        },
        "outputId": "4ed4b81a-60d1-4945-ee70-7c6529796aa5"
      },
      "source": [
        "class_labels = train_generator.class_indices\n",
        "class_labels"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'Angry': 0,\n",
              " 'Disgust': 1,\n",
              " 'Fear': 2,\n",
              " 'Happy': 3,\n",
              " 'Neutral': 4,\n",
              " 'Sad': 5,\n",
              " 'Suprise': 6}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NVTpY74MuARw",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "9758cd1b-0b28-4da0-a444-0076bfc51821"
      },
      "source": [
        "predict_generator = test_datagen.flow_from_directory(\n",
        "        '/content/Testing',\n",
        "        target_size=(48,48),\n",
        "        batch_size=1,\n",
        "        color_mode=\"grayscale\",\n",
        "        shuffle = False,\n",
        "        class_mode='categorical')"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 7178 images belonging to 7 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l316iEBardoo",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 89
        },
        "outputId": "5036447b-f6cc-469f-f4bb-67f5544a4ac6"
      },
      "source": [
        "Y_pred = model.predict_generator(predict_generator, nb_validation_samples)"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From <ipython-input-16-ab474448eb4c>:1: Model.predict_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use Model.predict, which supports generators.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UyY6a75Art53",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "y_pred = np.argmax(Y_pred,axis=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "icB_jOEaryPn",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "628bb58b-2998-4050-f593-94044d380600"
      },
      "source": [
        "len(y_pred)"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "7178"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oNjeNh4hr-RQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 262
        },
        "outputId": "a5b371a4-bb10-4b0a-d5a3-5500b67f08d7"
      },
      "source": [
        "print(classification_report(predict_generator.classes, y_pred))"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.46      0.54      0.50       958\n",
            "           1       0.66      0.21      0.32       111\n",
            "           2       0.48      0.23      0.31      1024\n",
            "           3       0.77      0.86      0.81      1774\n",
            "           4       0.53      0.56      0.54      1233\n",
            "           5       0.44      0.51      0.47      1247\n",
            "           6       0.74      0.71      0.72       831\n",
            "\n",
            "    accuracy                           0.59      7178\n",
            "   macro avg       0.58      0.52      0.52      7178\n",
            "weighted avg       0.58      0.59      0.57      7178\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7u4xJ50UFvcj",
        "colab_type": "text"
      },
      "source": [
        "#### The Highest Accuracy obtained using model as 59 %."
      ]
    }
  ]
}