Face_expression.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Face expression.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "mount_file_id": "1IKh9v95ypdeysTBju1lbuVOnZuIxIc2E",
      "authorship_tag": "ABX9TyPSCNUmla9maZcbRXhJ+jMx",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/melvin-02/Expression-detector/blob/master/Face_expression.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xU4CdCDmiJ3F",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "!unzip '/content/drive/My Drive/Colab Notebooks/expressions.zip'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KjAkhelAjcch",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "import matplotlib.pyplot as plt\n",
        "from glob import glob\n",
        "import os\n",
        "from tensorflow.keras.layers import Input, Dense, Conv2D, MaxPool2D, Dropout, Flatten, BatchNormalization\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator, load_img\n",
        "from tensorflow.keras.callbacks import ModelCheckpoint\n",
        "from tensorflow.keras.optimizers import Adam\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_kOau7-Cqfmv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "train_path = \"/content/images/train\"\n",
        "validation_path = \"/content/images/validation\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ro-Dr0b8q9wF",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "88cb1c63-1386-46ac-a9fc-e5b76d46cdaa"
      },
      "source": [
        "train_files = glob(train_path + '/*/*.jpg')\n",
        "valid_files = glob(validation_path + '/*/*.jpg')\n",
        "\n",
        "print(len(train_files), len(valid_files))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "28821 7066\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RjqipyCIrq7_",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "outputId": "821381b9-a67e-4f04-e57b-07bb2849a49c"
      },
      "source": [
        "img = load_img(train_files[0])\n",
        "plt.imshow(img)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fd17300dfd0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD6CAYAAABnLjEDAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO2de4yX1ZnHv08RBVHKRUAKOIJyKVaFhlqxm2Zr11bbWkzbbHrJxm1M7B+7jU27ae1ustkmu4n9p5dkN92Y7YVNmtqLJF7idoNIQ5pu0WmVKiAwgNwvTpEW0SrI2T/mN2be7/nO/A4D/Gbo+X4SI+dw3nPOe9738M7zned5TqSUYIz58+ctIz0BY0xn8GY3phK82Y2pBG92YyrBm92YSvBmN6YSzmizR8StEbElInoi4t6zNSljzNknhvt79ogYA2ArgFsA7AXwFIBPpZQ2DXbNZZddlrq6uhp1x44dG7IMAK+++mqjfPLkyazNqVOnGuW3vCX/d6zkXvk67hcAIqJtv9ymdPySvrluuGO1GxvQ96/Wtt346pqxY8cOeY2a0wUXXNB2Pur94LFUG9U3379aI1XHXHjhhY3yG2+80XasadOmte2HeeGFF9Db2ysnlN9dOTcA6Ekp7QCAiHgAwAoAg272rq4u/PrXv27UPfHEE43yunXrsus2btzYKB8+fDhr88orrzTKl1xySdbmxIkTjbJ6uSZMmNAov/zyy1mbMWPGNMrqwfHLVTo+P8zXX389a1PyAnIbNRZvEtWPGv+iiy5q2zffq3oeb3vb2xrlP/3pT1kbXo+pU6dmbfiZqfdj1qxZjXJvb2/W5rLLLsvqjh8/3ijzvQP5Oqp/NGbPnt0oq48aj/W5z30uazNnzpxGmZ/ZsmXLsmvenOegf9OeWQD2DCjvbdUZY0Yh51ygi4i7I6I7IrrVv6bGmM5wJpt9H4CBP1PMbtU1SCndn1JallJapn5MMsZ0hjOx2Z8CMD8i5qJvk38SwKeHuuD48eOZzf7kk09mbZijR482yn/4wx+yNhMnTmyUlSDE9laJXa0EEbbJlB2nbF22r1Tf3KZEEFI282uvvdYoq3vl8ZXNPG7cuKyuRHwbP3582zbcjxpLzZthO56foRpr8uTJWZsFCxZkddu3b2/bN+sB6n24+OKLG2VeH6BPXBvImjVrsjaf/exnG2UloA7GsDd7SulkRPw9gP8FMAbA91JKG9tcZowZIc7ky46U0mMAHjtLczHGnEPsQWdMJZzRl/10OXbsGNauXduo499ZHzhwILuOf4fO9g+Q20mqDdu6yrZiW1fZmvx7dTWWsu3YkUPZo9ymZHwF6wol2gM/C0Db0TxH9Xtlvjc1Z7ab+ffl6jplo/K9XX311Vkbvk7pEzNmzMjq+J3Zv39/1ob9BdR9sI2u1prfGeUvcOjQoUZZOd4Mhr/sxlSCN7sxleDNbkwleLMbUwkdFehOnDiBF198sVHHzjCTJk3Krrv00ksbZeVowtepqCYW0pRoxKKdElJ4fCXIqDly9J4an4UjJeKxg4gKMuH1UPPh8dW9lkRnlUSrKWGtRNhiByoWUIFc2HrrW9/adj7KqWbnzp1ZHQtgSnhlZxjlMMP3+q53vStrw89669atWRsOoFGi4mD4y25MJXizG1MJ3uzGVEJHbfZTp05lySDYllMOKlOmTGmUlaMHB9CUBJko26oEHl/pDMrWVffGlDjVcD/Tp0/P2rDdqgJzWENQNrsKTOJ1VH2rOobXTekTbKOzxgPkz6MkoEaN9dJLL2V1bMcrzaDE0YVtbX6ngVwvmj9/ftamJCvOYPjLbkwleLMbUwne7MZUgje7MZXQUYEupZQJUOyQwaIRkItUJVFmyvmBBSgVrcWinRqrJOOpEnJYXCrJwqIcVlh8U6IVr7OK8uI5qjkrsatEoOO+1H2U3D8/j5KMO+q5ljhUKUGsp6enUT548GDbOaq1ZhFPOcOweD137tyszcyZMxvlkozF/fjLbkwleLMbUwne7MZUQkdt9jFjxmRBG2xjKCcOvkbZluyQoLLQsG2nnGo4Uyif5AHkdqOymVXgB9uNyrZkG0zZw3wfyo5l7UM5Y7A9rmxN5WjDz0wFHfEzKgmoUf3w+MrJiO9NjcX9cGAKoN8HtptXr16dteGAFaXz8Hrs25dlXc9sds6qDOQZcEu0kX78ZTemErzZjakEb3ZjKsGb3ZhK6KhAFxGZcKaOcmI4gwmnlgZycUcJdOyQoKLVuI1yfmBBTEXYKdGMBSnVhkUyJZqxAKVEGRbflBjI/aj7UKIZ15Vk81GCZYnQx3WqHx6/JG22yu6j6lhsW7FiRdbmkUceaZR37NjRth8VYcfvmhLx+N1nAXeo46D8ZTemErzZjakEb3ZjKqHjgTDsBKDsPYbtNuU0wc4Wyv7q6upqlJXNzmfIq/mxra3s4ZL7UvYV34fquwS2h5VTDY9VYlervlSbkvvn60ruVT17fqdKHKrUnJWtz6igp9tvv71RfvDBB7M2W7ZsaZSV81hvb2+j/Mc//jFr8/zzzzfKfNTVUJls/GU3phK82Y2pBG92YyrBm92YSuioQAfkIkyJEwmLKSoLDTveqEg0TvGrsrDw/EpS96pILEWJsMVOEqpNyXhDRT/1w2KXcmopuX81Ftcp8a/kGKl21wC5IFdy1JUSyPiYMaDsnHnOHLRo0aKszd69exvl3//+91kbjnJTQiO/+yX33o+/7MZUgje7MZXQdrNHxPci4nBEPDegbkpErI6Iba3/50diGmNGFSU2+w8A/DuA/x5Qdy+ANSml+yLi3lb5K+06Uk41bIMoW4YdZFQmELa32DkGKMtCyjaPClYpsZmVrlByHTujKBuxxLZlSpxq1HooG1nNaTjjD0cPUY4vvNZK92E9otRZicdXa89rxMFUQH5E9O7du7M2nKlGaQjcD2fXGeoda/v2pZTWAThC1SsArGz9eSWAO9r1Y4wZWYZrs89IKR1o/fkggPIT4Y0xI8IZC3Sp7+eaQX+ujIi7I6I7IrrVj1fGmM4w3M1+KCJmAkDr/4cHa5hSuj+ltCyltEz9XtsY0xmG61TzMIA7AdzX+v9DJRedOnUqS3HMggKLFEAuyrATg4KdbBQlglSJo0lJtJS6rkTEU+OzQFYidJVEtJWe/c3tSsS/ElFR3Ss7GaksReyMUpL96Morr8zq1HFLnLr58ssvz9rweErUZeF1+vTpbftRQt9VV12V1ZVS8qu3HwH4PwALI2JvRNyFvk1+S0RsA/BXrbIxZhTT9sueUvrUIH/1/rM8F2PMOcQedMZUQsezy7Lt0u4IZ1VXctRvCcpmZvu75LihUocRtpvV0U6saZTYumqOvM4lGWDVWCWahRqf+yoJ+jl27FjWhusOHDiQteGgFnVs0pEjTVeRDRs2ZG1UJuHly5c3yh/84AezNhx0VeIIpY6f4mekjpBeuHBhozxUNlnGX3ZjKsGb3ZhK8GY3phK82Y2phI5nqmERgsUVlWGmJOVxiTNKSWYUdrRR4hP3rRw91PE+Jc4fPH7JUUZqjuytqNaV+1aC5ekIQANhkUqtEadKZhENyNdMZZhhoU+t2bx58xpldRa7qtu+fXujvGrVqqzNBz7wgayO4ftQTjVLly5tlEuy+5RmSQL8ZTemGrzZjakEb3ZjKsGb3ZhK6KhAN27cuOxsqk2bNjUnJCLRWDhSYg8Lf8o7rcTzi687ePBg1obrXnzxxayNit7jOSpBiFMRKRGR10MJOZz+S3n0cbqvkhTMaryS8ZV3HAty6mwzfh5qzTgKUoVSl6Qt47MAFSonAz9/lU6K+37hhReyNuwd19PTk7VhD0IVvTcY/rIbUwne7MZUgje7MZXQUZt94sSJuO222xp1hw83M1rt2bMnu67kKCG2CUtsTWWj8nx27NjRth8VLcX2l7quxKlH3QfXKWcYbsM2tBpLOaOo8Xk81YZtW6WzcJ3qh+dUcvY6p1sGctt///79becDANdff32jrLLHcN9qju9973sb5W3btmVtDh061ChPmjQpa8M6h212Y0yGN7sxleDNbkwleLMbUwkdFejGjx+Pt7/97Y06jv7ZunVrdh07KSiHGXaSUKmkOVpNRZ3t3LmzUWYnICA/510JMmqOjBLWOKqL01QBZZFP7HyixC928lGCoRIxeR2VYMpil1prFvFKzqJX6bVYfFQOK+2uAXSkIrd75zvfmbWZPXt2o6wcwzgN1ZIlS7I27DCjhN/hpBHvx192YyrBm92YSvBmN6YSRjxTzXAybyjbbvHixY2yStXLNuLevXuzNuw0oTKKsJ2knDFUwASfPa9SYrNNVnJsU4lTi3LQKDl7ryRtt7LHWWtQ+gQ7Qqn5sB6inhkH0HCAD5Db0eodUimoOaBJvZ+cBUi9ezz+O97xjqxNb29vo6z0CX6uJbpHP/6yG1MJ3uzGVII3uzGV4M1uTCV0XKBjWIBRAggLYEo04+gfdYY7i0YXX3xx1obP31YpqdnRgoW3werY0YYj7FTfSrRiZwvVhp1qlJMPOyupfpRTD4ttShRiYUuJXywuKUGKs8CoDEB8rroSRxmVFackSxJnVgLyM9OVQMeReerdUyImw9FyLLwOde/+shtTCd7sxlSCN7sxlTDi57MrB4h2qGwhbG8qZxS2yTigBchtK2Wzsj2uHD2Uow07kagAGrbZ1fhsby5YsCBrU3KuO9+/stlVHa+tcurhgB51bBPPUWWg5Xu96aabsjZsxz711FNZGw5ouu6667I2KgiL7Whla7Ouo4JT+F7VurY7Gg3INa3HHnusUVYOTm9eO+jfGGP+rPBmN6YSvNmNqYS2mz0i5kTE2ojYFBEbI+KeVv2UiFgdEdta/89TehpjRg0lAt1JAF9KKf02Ii4F8JuIWA3gbwGsSSndFxH3ArgXwFfaddYus0bJMUVK2CpJwcziW0m0mHK+4HTXSqBTxwux88ejjz6atWGxh1MQqzacghjInYxUmmgWe0qEJaAsgozXVjlL8XUqe8ysWbMaZSVs8fNQzilbtmxplPksdABYvnx5VsdZb5RAx/ehHJhKRE1GOflwP5zqXK1hP22/7CmlAyml37b+fAzAZgCzAKwAsLLVbCWAO9r1ZYwZOU7LZo+IKwEsBbAewIyUUn/SrIMA8oRZfdfcHRHdEdGtDkA0xnSG4s0eEZcAeBDAF1JKjZ9tU9/PMfIXuyml+1NKy1JKy9TvtY0xnaHIqSYixqJvo/8wpbSqVX0oImamlA5ExEwAeVQH8frrr2f2FTsOKJuM7RsVnMI2obIj2WlB2bF8nXKO4Z9QSo77AYBVq1Y1ysqJhG0wdUTx7bff3iiroBvOeKqOEWbbtmRdFeq6kn74WavnwW2UHcvrr54HO2KVZAACgM9//vONsrLH2U5WmY25bzVH1ovU+8GZg+bNm9e2335K1PgA8F0Am1NK3xjwVw8DuLP15zsBPNSuL2PMyFHyZX8PgL8B8GxEPNOq+0cA9wH4SUTcBWAXgL8+N1M0xpwN2m72lNIvAQz2+7L3n93pGGPOFfagM6YSOhr1dvz4cfzqV79q1LH4xml5gfxYHhUJxiKRynpSctxRSWprdo5RY5WkYObjfoA8w87UqVOzNiyslWRGUQ4zLH6pNiXOSWqOfMzX888/n7VhZyAVAclrq8TI973vfY1yyXFUSlRUYihnxlFHMrFoqBxbWDhTGWX4eaijplj84/VZuXIlBsNfdmMqwZvdmErwZjemEjpqs7/yyit49tlnG3XDOWpZZRhl21LZ0Wx/lziMKHv4iiuuaJRVdhAVnMK2JTu+APl9qMALtklV4AfbhMphhW1NZeuqNWKbVLW58cYbG2W1jg8//HDb8UsCSNjRRb1D7Y4dU/0AeZCT0nn4HS7JQKT0EZ5Td3d31oazEnEGnqECzfxlN6YSvNmNqQRvdmMqwZvdmEroqEA3YcKEzFGAxQ3l2MDZQpQzCosryqmFhb6SCCol9jDKGUSJZuwMtGjRoqwNH5OkRCMWiVREGzvnqHVllGilHERYkFNHbfF1KnUzj7dmzZqsDQuL6rmWCJZ8nWqjHG1YfFVOX+r+26HeK3b82bx5c9aGIy75Pnz8kzHGm92YWvBmN6YSvNmNqYSOCnRjx47NvMbYQ2zhwoXZdSw69Pb2tm1TknZICXQsPimBiq8rTefEXlzK02ry5Gb6fXXeF0dHcbpl1UZ5HbL4VXKuuUKJiFynPM/4WauU3Lt3726U1TNj8a0kTbN6PxRDpWbuh9etJP22EvW6uroaZeWFyfMuWed+/GU3phK82Y2pBG92Yyqhozb7yZMnM3t70qRJjbKyZdjOZzsOyJ1RVEYRtpGVrcn2XomDRmmGl5Ijkdi2mz59etZmypQpjXKJM4xy/GEnn5IjmoB8TdS9skZQ0vf111+ftWGHoZ07d2Zt+LmqOZecj14SGai0D87Uo+615L3i97ynpydrw/oQO+cMFcnpL7sxleDNbkwleLMbUwne7MZUQkcFOiAXkzga68iRI9k1P//5zxtllaqJUQIIC3LswKJQ4leJ2KKcMbgv5SDCdSpNMztxlJyjpkQ0PnterasSH1Uqb4YFU45cBPKUSuvXr8/asAClHIg4vbQS2rgfdQ8qlRkLxmqtWXRm5xggP7dNRSqy+KfSnfE7WyIW9+MvuzGV4M1uTCV4sxtTCR212SMis6fYRn/uueey69j+vfbaa7M211xzTaOsbFS2Z0qCTFQ/bLcpO6kknbGy2dlOU0E2JZlZeE4qyIXvTc1HXcf3ppyT2GZXDiKsEbCDFZDb8cpm53VV61GyZiUpoNVz5b6Vrc1OXnykGZCvGb/TQK5x8X04EMYY481uTC14sxtTCd7sxlRCRwW6U6dOZelyWRRR52+zY4lKS8xRXSo6icULJX6xk4ISn0rOSFMOKjy+ElP4OnbGAHIhi9cUyIUbNccSRxM1Pq+JcpjhlMfquW7YsKFRVs+V72Pbtm1ZG05Hrp4ZP9fSZ8Z9lURBqndv2rRpjbISPnmtP/7xj2dt1q5d2yjzMxsqA4+/7MZUgje7MZXQdrNHxLiIeDIiNkTExoj4Wqt+bkSsj4ieiPhxROQOycaYUUOJzf4agJtTSi9HxFgAv4yI/wHwRQDfTCk9EBH/CeAuAN8ZqqOIyOxddv5Qjg1sk6mABbatlKMLO0iosdiOVvZfSSZbZdupebebo8qown0r5yC2I9WxSXwfjz/+eNZGOcMsX768UVbZftlmX7duXdZGaQ0Mr5kKMOKxlO1dkhFYPTPuqySgSGUX4mdU4iyl9Al+H/g4taGOK2v7ZU999CsOY1v/JQA3A/hZq34lgDva9WWMGTmKbPaIGBMRzwA4DGA1gO0AjqaU+v952gsg92M0xowaijZ7SumNlNISALMB3AAgP350ECLi7ojojohu5TNsjOkMp6XGp5SOAlgLYDmASRHRbwjNBrBvkGvuTyktSyktG87RtsaYs0NbgS4ipgE4kVI6GhHjAdwC4Ovo2/SfAPAAgDsBPNSur5RSJnCwADN37tzsOnbIUIIUCzdKAGFxQwk5fN1QqXn7UWefKwcVFomUUw0LaSrrSkmkE9cpYank3HuVgpqzpagILnYQKXGE2rhxY9aGn4cSQ1loVc+Vn4daV+VAxKi1Zicn9c6wo4368LHwqe7jqquuapSvuOKKRlmJzv2UqPEzAayMiDHo+0ngJymlRyNiE4AHIuJfATwN4LsFfRljRoi2mz2l9DsAS0X9DvTZ78aY8wB70BlTCR0NhBk/fjwWL17cqGPHAT4CB8iDCJS9dfDgwUZ54sSJWZuS459YQ1A2K9uRyllG2VtsbypHF0YFTLDdqDQDrlN2JK/1xz72sayNyl7D2oeyo/k65RzE17FzjOpH3Ss/I6XX8LNWc1YZX3ne6pnx81DOQvwO8/sK5EdbzZs3L2vDR2S1c1IbiL/sxlSCN7sxleDNbkwleLMbUwkdFeguuuiizCmA0+fu3bs3u06JbcycOXMaZSUssTNK6bnqDItdShAqSUFdEvk0VOaRfoYSZfpRziB8zvtQaYgHwmKXivJi5ygeCwB27drVKO/Zsydrc+jQoUZZCZYcdaZEPL5OibzKWYufmXKGUUIvU5K2mwW5pUuz33jL60rxl92YSvBmN6YSvNmNqYSO2uwppcwGXbJkSaOsbFQOjVUOGiVHMnGbEru6pI2yddX4fG/DPVqKUfYnO/oofaBEQygZXznscPCSclhhR5OSI5MPHz6cteGAnpIMsCXHgwH5u6Yy5fDzV3Y1981BL6XwWCXPpx9/2Y2pBG92YyrBm92YSvBmN6YSOn4+O4sXLJSwYAcAW7ZsaZRVZpSSjCbs/KCyerBIpAQQ7qckEgrIhSMl5JQcEcUoYYvHUm1KxEAl2rFIpZxKWFRlxxcgd35RzjCMWjN2vFHRcxwZp0RFtUbsjFOSXUjdh3IqYngdlfB6OoIc4y+7MZXgzW5MJXizG1MJHbXZFWyDKDvlmmuuaZTV8b/79jUzWSub/fLLL2+Ud+/enbXhDKPK/uK+lROHsu34OmU3so2s+uHrVD88lroPtr2Vo0mJ9qHun8crOVZaBZmwU4sai98h5fjCa6Rs/9I6hrUfZfuzA5Gi5HiwdtmZh8JfdmMqwZvdmErwZjemErzZjamEUSfQlUSr8XntQJ4tRQlL3LcSe9hhRwlLLIqofhR8HyrrCs+xJFONgueoxDc+kkjNRx3GyXNSbTjrDI8F5Om2Fy3KzwtduHBho8xHLQFlZ5/zMxpu1FuJkxMfyQTkwrMan5+ZGqukzWD4y25MJXizG1MJ3uzGVII3uzGVMOoEupIUT0pIYa8pFdHGfbNHHZCfUa4iqFhsUSJaybnuJWmgVEQZi02qnwkTJjTKSmhkYU0JbUp85PTfKlXU9u3bG+WSc9X379+ftdmxY0ejPH/+/KyNEhaZkvdM3WuJQMprPWPGjKxNSSozpiS1Wcl79mbb4pbGmPMab3ZjKsGb3ZhK6LjNPpxUuCV2Ctt/JU4LKiU12/Eqwo77LnV84XsvyYyi+uZ+lM3ONjqnWwZyW5szvgB5NCGQr5vSFdj+VpoBz1tl/Onu7m6U+fgwAJg8eXKjrN4pXjM1lqrjvtQ7w+fcz5w5M2tT8g6XtOG603G68pfdmErwZjemEoo3e0SMiYinI+LRVnluRKyPiJ6I+HFE5FknjDGjhtP5st8DYPOA8tcBfDOldDWAlwDcdTYnZow5uxQJdBExG8CHAfwbgC9Gn2pxM4BPt5qsBPAvAL5T0Ndp//1w0ueWpOtR4gZHLKnz4tmpRKGcJrhOCTDssKMcRkrOkN+0aVOj/P3vfz9rw+mdVTontfYla8sRbWo9eDy1HiWps/gccxbsgHzOSlRUjlh8Hyp1FqdNU4Ipp7cariNWJ9JSfQvAlwH0z3AqgKMppf672gtgVvGoxpiO03azR8RHABxOKf1mOANExN0R0R0R3cr11BjTGUq+7O8B8NGIeAHAA+j78f3bACZFRP/PJrMB5L+QBZBSuj+ltCyltKwkw6Yx5tzQ1mZPKX0VwFcBICL+EsA/pJQ+ExE/BfAJ9P0DcCeAh87hPM8JykZi27Krqytrww4iyv5TsN1WctySsiN5fBXA8sQTTzTKnDkGyM9MnzhxYtamJKBIwdoDB4sAefYa5cDENrOydfk+1JxVinJGOczwsVEqCw3b8SXpp4fLSB3/9BX0iXU96LPhv3sGfRljzjGn9U9QSukXAH7R+vMOADec/SkZY84F9qAzphK82Y2phBHPVDPamTUrdx84ePBgo6wytZQ4OyiBkCOv+Ow5IBfIlCDEzigqoqvk/DN1RlyJ0MiCnFoPdhBRAhmLn+o3OiyiKQGxJCW0Eig5bfmCBQuyNiXi33CiPRU+n90Y0xZvdmMqwZvdmEqwzU6w04ayNTkziXJqUbYVZy8tsetVGz7uiOcDANdee22j/PTTT7ftW82Z7WEgt+NfffXVrI3KsMooPYJhrUEdEVWSkZfbKPtcBdDwsWJKM2B7fLgZkpnhBssMhr/sxlSCN7sxleDNbkwleLMbUwkW6IiS869ZtNm1a1fWhkU01XeJSKMEOr5ORXldd911jbKK1uL8Aursc9U3O8yo61jIUmJcyVqz0KfEyCNHjjTKJdmOhivQKeekdmMNVscMx/HG57MbYzK82Y2pBG92Yyqhapu9xPlBwYEgnCkF0DYq23unY28NhDOjlOgKN910U9bmkUceaZSVM4rSDPj+OVgEKAuy4XkrnYPvQ2Vu5bVWz4PvQ9ns6jplxzNnEpzSyX79ZTemErzZjakEb3ZjKsGb3ZhKqFqgKxFAStooYUcJUixaqaOMuE5liuHouZLoKOWMsnjx4ka5p6cna8MOK0B+v2qNWGxTgiVn+FFiGEe5qQg7Rq0HOwepY5w4bTWQZ7gZrmhWIsYOp+/TucZfdmMqwZvdmErwZjemEqq22RVsM6vMIGwnKVtPBZCUHBPFtr6y60uCMdhpRAXCvPvd726UDxw4kLXhY50BYN++5rF+yhmG561sf86Cc+utt2ZtSoJlWENQjjdse5cedVUCawQl70wJZ9vO95fdmErwZjemErzZjakEb3ZjKiGGG3k1rMEiXgSwC8BlAHo7NvDZ4XycM3B+zttzHj5dKaU83zU6vNnfHDSiO6W0rOMDnwHn45yB83PenvO5wT/GG1MJ3uzGVMJIbfb7R2jcM+F8nDNwfs7bcz4HjIjNbozpPP4x3phK6Phmj4hbI2JLRPRExL2dHr+EiPheRByOiOcG1E2JiNURsa31//aZCDtIRMyJiLURsSkiNkbEPa36UTvviBgXEU9GxIbWnL/Wqp8bEetb78iPI+LCkZ4rExFjIuLpiHi0VR71c+7oZo+IMQD+A8BtABYD+FRELB76qhHhBwA4KuNeAGtSSvMBrGmVRxMnAXwppbQYwI0A/q61tqN53q8BuDmldD2AJQBujYgbAXwdwDdTSlcDeAnAXSM4x8G4B8DmAeVRP+dOf9lvANCTUtqRUnodwAMAVnR4Dm1JKa0DwGFaKwCsbP15JYA7OjqpNqSUDqSUftv68zH0vYizMIrnnfp4uVUc23pruBwAAAHdSURBVPovAbgZwM9a9aNqzgAQEbMBfBjAf7XKgVE+Z6Dzm30WgD0DyntbdecDM1JK/TGgBwHMGKrxSBIRVwJYCmA9Rvm8Wz8OPwPgMIDVALYDOJpS6o9THY3vyLcAfBlAf2zrVIz+OVugGw6p71cYo/LXGBFxCYAHAXwhpdQIRh+N804pvZFSWgJgNvp+8lvU5pIRJSI+AuBwSuk3Iz2X06XTySv2AZgzoDy7VXc+cCgiZqaUDkTETPR9iUYVETEWfRv9hymlVa3qUT9vAEgpHY2ItQCWA5gUERe0vpSj7R15D4CPRsSHAIwDMBHAtzG65wyg81/2pwDMbymXFwL4JICHOzyH4fIwgDtbf74TwEMjOJeMlt34XQCbU0rfGPBXo3beETEtIia1/jwewC3o0xrWAvhEq9momnNK6asppdkppSvR9/4+kVL6DEbxnN8kpdTR/wB8CMBW9Nlm/9Tp8Qvn+CMABwCcQJ/9dRf67LI1ALYBeBzAlJGeJ835L9D3I/rvADzT+u9Do3neAK4D8HRrzs8B+OdW/TwATwLoAfBTABeN9FwHmf9fAnj0fJmzPeiMqQQLdMZUgje7MZXgzW5MJXizG1MJ3uzGVII3uzGV4M1uTCV4sxtTCf8PWo9xUjEXf9wAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kWEyvVSpsA0N",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "i = Input(shape = (48,48,1))\n",
        "x = Conv2D(32, (3,3), padding='same', activation='relu')(i)\n",
        "x = Conv2D(32, (3,3), padding='same', activation='relu')(x)\n",
        "x = MaxPool2D((2,2) )(x)\n",
        "x = Conv2D(64, (5,5), padding='same', activation='relu')(x)\n",
        "x = Conv2D(64, (3,3), padding='same', activation='relu')(x)\n",
        "x = MaxPool2D((2,2))(x)\n",
        "x = Dropout(0.5)(x)\n",
        "x = Conv2D(128, (3,3), padding='same', activation='relu')(x)\n",
        "x = Conv2D(128, (3,3), padding='same', activation='relu')(x)\n",
        "x = MaxPool2D((2,2))(x)\n",
        "x = Dropout(0.5)(x)\n",
        "x = Flatten()(x)\n",
        "x = Dense(256, activation='relu')(x)\n",
        "x = Dropout(0.5)(x)\n",
        "x = Dense(7, activation='softmax')(x)\n",
        "\n",
        "model = Model(i, x)\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kThHKobgwA6W",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.compile(optimizer= Adam(learning_rate=0.001), loss='categorical_crossentropy', metrics=['accuracy'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZkOKRMWixFho",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "train_generator = ImageDataGenerator(rescale=1/255.0) \n",
        "                              \n",
        "\n",
        "test_generator = ImageDataGenerator(rescale=1/255.0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "getIJny7xtRC",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "1e845dde-7315-4cf5-d9b8-f975b5ce92d8"
      },
      "source": [
        "train_gen = train_generator.flow_from_directory(train_path,\n",
        "                                          target_size = (48,48),\n",
        "                                          color_mode = \"grayscale\",\n",
        "                                          class_mode = 'categorical',\n",
        "                                          shuffle = True,\n",
        "                                          batch_size = 64)\n",
        "\n",
        "test_gen = test_generator.flow_from_directory(validation_path,\n",
        "                                         target_size = (48,48),\n",
        "                                         color_mode = 'grayscale',\n",
        "                                         class_mode = 'categorical',\n",
        "                                         shuffle = False,\n",
        "                                         batch_size = 64)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 28821 images belonging to 7 classes.\n",
            "Found 7066 images belonging to 7 classes.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "grsH9Y612D9B",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# save only the best model with the least loss\n",
        "model_checkpoint_callback = ModelCheckpoint(\"/content/faceExp.h5\",\n",
        "                                            monitor = 'val_loss',\n",
        "                                            save_best_only = True,\n",
        "                                            save_weights_only  = False,\n",
        "                                            )"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yoNsDALFym_S",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 938
        },
        "outputId": "b6c07601-3948-4a1d-8ca7-b28bc8b1939c"
      },
      "source": [
        "r = model.fit_generator(train_gen,\n",
        "                    steps_per_epoch = len(train_gen.filenames)//64,\n",
        "                    epochs = 25,\n",
        "                    validation_data = test_gen,\n",
        "                    validation_steps = len(test_gen)//64,\n",
        "                    workers = -1,\n",
        "                    callbacks = [model_checkpoint_callback])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From <ipython-input-11-e309b2357a0c>: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 1/25\n",
            "450/450 [==============================] - 13s 29ms/step - loss: 1.8082 - accuracy: 0.2459 - val_loss: 2.0569 - val_accuracy: 0.0000e+00\n",
            "Epoch 2/25\n",
            "450/450 [==============================] - 13s 28ms/step - loss: 1.6416 - accuracy: 0.3485 - val_loss: 1.6946 - val_accuracy: 0.1094\n",
            "Epoch 3/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.4618 - accuracy: 0.4315 - val_loss: 1.6770 - val_accuracy: 0.2031\n",
            "Epoch 4/25\n",
            "450/450 [==============================] - 12s 28ms/step - loss: 1.3681 - accuracy: 0.4725 - val_loss: 1.4730 - val_accuracy: 0.3750\n",
            "Epoch 5/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.3063 - accuracy: 0.4992 - val_loss: 1.6144 - val_accuracy: 0.3594\n",
            "Epoch 6/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.2581 - accuracy: 0.5207 - val_loss: 1.2928 - val_accuracy: 0.5312\n",
            "Epoch 7/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.2307 - accuracy: 0.5303 - val_loss: 1.4641 - val_accuracy: 0.5156\n",
            "Epoch 8/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.2019 - accuracy: 0.5408 - val_loss: 1.5162 - val_accuracy: 0.3906\n",
            "Epoch 9/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.1797 - accuracy: 0.5494 - val_loss: 1.3929 - val_accuracy: 0.4219\n",
            "Epoch 10/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.1630 - accuracy: 0.5552 - val_loss: 1.3309 - val_accuracy: 0.5000\n",
            "Epoch 11/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.1415 - accuracy: 0.5651 - val_loss: 1.2733 - val_accuracy: 0.4844\n",
            "Epoch 12/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.1267 - accuracy: 0.5733 - val_loss: 1.3892 - val_accuracy: 0.4688\n",
            "Epoch 13/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.1138 - accuracy: 0.5761 - val_loss: 1.2178 - val_accuracy: 0.5781\n",
            "Epoch 14/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.0999 - accuracy: 0.5803 - val_loss: 1.3044 - val_accuracy: 0.5156\n",
            "Epoch 15/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.0803 - accuracy: 0.5870 - val_loss: 1.3241 - val_accuracy: 0.4844\n",
            "Epoch 16/25\n",
            "450/450 [==============================] - 12s 28ms/step - loss: 1.0781 - accuracy: 0.5892 - val_loss: 1.2121 - val_accuracy: 0.5625\n",
            "Epoch 17/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.0608 - accuracy: 0.5951 - val_loss: 1.2543 - val_accuracy: 0.5156\n",
            "Epoch 18/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.0512 - accuracy: 0.5970 - val_loss: 1.1683 - val_accuracy: 0.5625\n",
            "Epoch 19/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.0408 - accuracy: 0.6054 - val_loss: 1.2499 - val_accuracy: 0.4844\n",
            "Epoch 20/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.0336 - accuracy: 0.6077 - val_loss: 1.2144 - val_accuracy: 0.5156\n",
            "Epoch 21/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.0297 - accuracy: 0.6085 - val_loss: 1.0460 - val_accuracy: 0.6250\n",
            "Epoch 22/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.0152 - accuracy: 0.6111 - val_loss: 1.0577 - val_accuracy: 0.6250\n",
            "Epoch 23/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 1.0072 - accuracy: 0.6186 - val_loss: 1.1687 - val_accuracy: 0.5000\n",
            "Epoch 24/25\n",
            "450/450 [==============================] - 12s 27ms/step - loss: 0.9995 - accuracy: 0.6194 - val_loss: 1.2196 - val_accuracy: 0.5469\n",
            "Epoch 25/25\n",
            "450/450 [==============================] - 13s 28ms/step - loss: 0.9916 - accuracy: 0.6217 - val_loss: 1.2669 - val_accuracy: 0.5156\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z2nYVcRY0Pas",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "286995d7-abc0-443c-e881-8d1c66c65b08"
      },
      "source": [
        "plt.plot(r.history['loss'], label='loss')\n",
        "plt.plot(r.history['val_loss'], label='val_loss')\n",
        "plt.legend()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7fd10f53e160>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        },
        {
          "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": "OqX4HBuWS-NV",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "6fbcda54-6469-4007-b5b8-4149926e6ec7"
      },
      "source": [
        "plt.plot(r.history['accuracy'], label='accuracy')\n",
        "plt.plot(r.history['val_accuracy'], label='val_accuracy')\n",
        "plt.legend()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7fd10f4e2978>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        },
        {
          "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": "3tdqG4UXDRIa",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 105
        },
        "outputId": "d584f1c5-95e1-4d2b-ade4-2fc02bddc268"
      },
      "source": [
        "model.evaluate_generator(test_gen)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From <ipython-input-14-162faf28a7af>:1: Model.evaluate_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.evaluate, which supports generators.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[1.0812371969223022, 0.606000542640686]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "z3sJ5MOmVqe4",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "callback_model = tf.keras.models.load_model('/content/faceExp.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8q8cem-VWBOU",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "84ebf2a7-5f23-4a6c-b141-42207a409a13"
      },
      "source": [
        "callback_model.evaluate_generator(test_gen)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[1.0742440223693848, 0.6016133427619934]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5wAYO6bTEU8h",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.save('/content/drive/My Drive/FaceExp60.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qHyRadNvobQd",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "saved_model = tf.keras.models.load_model('/content/drive/My Drive/FaceExp61.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LJfydF2BoiMy",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "d4dc12a8-ba25-42e7-8a73-f853cd3719ab"
      },
      "source": [
        "saved_model.evaluate_generator(test_gen)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[1.0546190738677979, 0.6140673756599426]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "A_qzHVvPTJ8j",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "label = 'angry disgust fear happy neutral sad surprise'.split()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pZwJ6pNEEy0Y",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "outputId": "f8542f24-0c44-48d9-d164-493a83ca4ebd"
      },
      "source": [
        "img = load_img(train_files[0], grayscale=True)\n",
        "img2 = tf.keras.preprocessing.image.img_to_array(img)\n",
        "img2 = np.expand_dims(img2, axis=0)\n",
        "model.predict(img2)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/keras_preprocessing/image/utils.py:107: UserWarning: grayscale is deprecated. Please use color_mode = \"grayscale\"\n",
            "  warnings.warn('grayscale is deprecated. Please use '\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[0., 0., 1., 0., 0., 0., 0.]], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 92
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fpxx3MhmeRwx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vaVQNQWTe7E5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}