facialKeyPointsDetection.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "facialKeyPointsDetection.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "5FVQbd14GOlr",
        "outputId": "89649475-c483-46ba-a40e-e79c26c37570",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "# this cell is only for colab to download the dataset\n",
        "!pip install -q kaggle\n",
        "!mkdir -p ~/.kaggle\n",
        "!cp kaggle.json ~/.kaggle/\n",
        "!chmod 600 /root/.kaggle/kaggle.json\n",
        "!kaggle competitions download -c facial-keypoints-detection"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Warning: Looks like you're using an outdated API Version, please consider updating (server 1.5.9 / client 1.5.4)\n",
            "Downloading IdLookupTable.csv to /content\n",
            "  0% 0.00/843k [00:00<?, ?B/s]\n",
            "100% 843k/843k [00:00<00:00, 56.8MB/s]\n",
            "Downloading training.zip to /content\n",
            " 82% 49.0M/60.1M [00:01<00:00, 39.9MB/s]\n",
            "100% 60.1M/60.1M [00:01<00:00, 56.5MB/s]\n",
            "Downloading SampleSubmission.csv to /content\n",
            "  0% 0.00/201k [00:00<?, ?B/s]\n",
            "100% 201k/201k [00:00<00:00, 68.2MB/s]\n",
            "Downloading test.zip to /content\n",
            " 31% 5.00M/16.0M [00:00<00:00, 40.3MB/s]\n",
            "100% 16.0M/16.0M [00:00<00:00, 78.3MB/s]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lY0CaXcJGYfg",
        "outputId": "aa9ed3e5-040b-4c53-bfee-3f2c866eb895",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "# This code is only for colab to unzip the dataset in colab content directory\n",
        "!unzip training.zip -d train\n",
        "!unzip test.zip -d test"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Archive:  training.zip\n",
            "  inflating: train/training.csv      \n",
            "Archive:  test.zip\n",
            "  inflating: test/test.csv           \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "v0y4XdXr4nUD"
      },
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "from matplotlib import pyplot as plt\n",
        "import seaborn as sns\n",
        "%matplotlib inline"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H2gkmW2DIvyh"
      },
      "source": [
        "from tqdm import tqdm"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CZ0EUHrjIvvW"
      },
      "source": [
        "from keras.layers.advanced_activations import LeakyReLU\n",
        "from tensorflow.keras.models import Sequential, Model\n",
        "from tensorflow.keras.layers import Activation, Convolution2D, MaxPooling2D, BatchNormalization, Flatten\n",
        "from tensorflow.keras.layers import Dense, Dropout, Conv2D, MaxPool2D, ZeroPadding2D"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ssw4bo1E5Ujb"
      },
      "source": [
        "train = pd.read_csv(\"train/training.csv\")\n",
        "test = pd.read_csv(\"test/test.csv\")\n",
        "lookup = pd.read_csv(\"IdLookupTable.csv\")\n",
        "sample = pd.read_csv(\"SampleSubmission.csv\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7k3Elfa47lRY",
        "outputId": "7288eb4c-0f2f-41e2-ff60-f1bdb55a9867",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "train.isnull().sum()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "left_eye_center_x              10\n",
              "left_eye_center_y              10\n",
              "right_eye_center_x             13\n",
              "right_eye_center_y             13\n",
              "left_eye_inner_corner_x      4778\n",
              "left_eye_inner_corner_y      4778\n",
              "left_eye_outer_corner_x      4782\n",
              "left_eye_outer_corner_y      4782\n",
              "right_eye_inner_corner_x     4781\n",
              "right_eye_inner_corner_y     4781\n",
              "right_eye_outer_corner_x     4781\n",
              "right_eye_outer_corner_y     4781\n",
              "left_eyebrow_inner_end_x     4779\n",
              "left_eyebrow_inner_end_y     4779\n",
              "left_eyebrow_outer_end_x     4824\n",
              "left_eyebrow_outer_end_y     4824\n",
              "right_eyebrow_inner_end_x    4779\n",
              "right_eyebrow_inner_end_y    4779\n",
              "right_eyebrow_outer_end_x    4813\n",
              "right_eyebrow_outer_end_y    4813\n",
              "nose_tip_x                      0\n",
              "nose_tip_y                      0\n",
              "mouth_left_corner_x          4780\n",
              "mouth_left_corner_y          4780\n",
              "mouth_right_corner_x         4779\n",
              "mouth_right_corner_y         4779\n",
              "mouth_center_top_lip_x       4774\n",
              "mouth_center_top_lip_y       4774\n",
              "mouth_center_bottom_lip_x      33\n",
              "mouth_center_bottom_lip_y      33\n",
              "Image                           0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sOOZqn5-H3-a"
      },
      "source": [
        "train.fillna(method='ffill', inplace=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Y4JXPng-I7cT"
      },
      "source": [
        "imag = []\n",
        "for i in range(0, 7049):\n",
        "    img = train[\"Image\"][i].split(' ')\n",
        "    img = ['0' if x == '' else x for x in img]\n",
        "    imag.append(img)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NzSqfokNPI1t"
      },
      "source": [
        "image_list = np.array(imag, dtype=\"float\")\n",
        "X_train = image_list.reshape(-1, 96, 96, 1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s3EnPmJCPmy4",
        "outputId": "c824cdb1-445b-4d4f-fe2c-e4cb4cfa947c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        }
      },
      "source": [
        "plt.imshow(X_train[5].reshape(96,96),cmap='gray')\n",
        "plt.show()"
      ],
      "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"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JYMSpk2fP3jg"
      },
      "source": [
        "training = train.drop('Image',axis = 1)\n",
        "\n",
        "y_train = []\n",
        "for i in range(0,7049):\n",
        "    y = training.iloc[i,:]\n",
        "\n",
        "    y_train.append(y)\n",
        "y_train = np.array(y_train,dtype = 'float')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w_n3KBBbXB8W",
        "outputId": "ac46dc9c-6da7-4d1d-99d8-2031ef07862d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 224
        }
      },
      "source": [
        "y_train = pd.DataFrame(y_train)\n",
        "y_train.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "      <th>2</th>\n",
              "      <th>3</th>\n",
              "      <th>4</th>\n",
              "      <th>5</th>\n",
              "      <th>6</th>\n",
              "      <th>7</th>\n",
              "      <th>8</th>\n",
              "      <th>9</th>\n",
              "      <th>10</th>\n",
              "      <th>11</th>\n",
              "      <th>12</th>\n",
              "      <th>13</th>\n",
              "      <th>14</th>\n",
              "      <th>15</th>\n",
              "      <th>16</th>\n",
              "      <th>17</th>\n",
              "      <th>18</th>\n",
              "      <th>19</th>\n",
              "      <th>20</th>\n",
              "      <th>21</th>\n",
              "      <th>22</th>\n",
              "      <th>23</th>\n",
              "      <th>24</th>\n",
              "      <th>25</th>\n",
              "      <th>26</th>\n",
              "      <th>27</th>\n",
              "      <th>28</th>\n",
              "      <th>29</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>66.033564</td>\n",
              "      <td>39.002274</td>\n",
              "      <td>30.227008</td>\n",
              "      <td>36.421678</td>\n",
              "      <td>59.582075</td>\n",
              "      <td>39.647423</td>\n",
              "      <td>73.130346</td>\n",
              "      <td>39.969997</td>\n",
              "      <td>36.356571</td>\n",
              "      <td>37.389402</td>\n",
              "      <td>23.452872</td>\n",
              "      <td>37.389402</td>\n",
              "      <td>56.953263</td>\n",
              "      <td>29.033648</td>\n",
              "      <td>80.227128</td>\n",
              "      <td>32.228138</td>\n",
              "      <td>40.227609</td>\n",
              "      <td>29.002322</td>\n",
              "      <td>16.356379</td>\n",
              "      <td>29.647471</td>\n",
              "      <td>44.420571</td>\n",
              "      <td>57.066803</td>\n",
              "      <td>61.195308</td>\n",
              "      <td>79.970165</td>\n",
              "      <td>28.614496</td>\n",
              "      <td>77.388992</td>\n",
              "      <td>43.312602</td>\n",
              "      <td>72.935459</td>\n",
              "      <td>43.130707</td>\n",
              "      <td>84.485774</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>64.332936</td>\n",
              "      <td>34.970077</td>\n",
              "      <td>29.949277</td>\n",
              "      <td>33.448715</td>\n",
              "      <td>58.856170</td>\n",
              "      <td>35.274349</td>\n",
              "      <td>70.722723</td>\n",
              "      <td>36.187166</td>\n",
              "      <td>36.034723</td>\n",
              "      <td>34.361532</td>\n",
              "      <td>24.472511</td>\n",
              "      <td>33.144443</td>\n",
              "      <td>53.987404</td>\n",
              "      <td>28.275949</td>\n",
              "      <td>78.634213</td>\n",
              "      <td>30.405923</td>\n",
              "      <td>42.728851</td>\n",
              "      <td>26.146043</td>\n",
              "      <td>16.865362</td>\n",
              "      <td>27.058860</td>\n",
              "      <td>48.206298</td>\n",
              "      <td>55.660936</td>\n",
              "      <td>56.421447</td>\n",
              "      <td>76.352000</td>\n",
              "      <td>35.122383</td>\n",
              "      <td>76.047660</td>\n",
              "      <td>46.684596</td>\n",
              "      <td>70.266553</td>\n",
              "      <td>45.467915</td>\n",
              "      <td>85.480170</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>65.057053</td>\n",
              "      <td>34.909642</td>\n",
              "      <td>30.903789</td>\n",
              "      <td>34.909642</td>\n",
              "      <td>59.412000</td>\n",
              "      <td>36.320968</td>\n",
              "      <td>70.984421</td>\n",
              "      <td>36.320968</td>\n",
              "      <td>37.678105</td>\n",
              "      <td>36.320968</td>\n",
              "      <td>24.976421</td>\n",
              "      <td>36.603221</td>\n",
              "      <td>55.742526</td>\n",
              "      <td>27.570947</td>\n",
              "      <td>78.887368</td>\n",
              "      <td>32.651621</td>\n",
              "      <td>42.193895</td>\n",
              "      <td>28.135453</td>\n",
              "      <td>16.791158</td>\n",
              "      <td>32.087116</td>\n",
              "      <td>47.557263</td>\n",
              "      <td>53.538947</td>\n",
              "      <td>60.822947</td>\n",
              "      <td>73.014316</td>\n",
              "      <td>33.726316</td>\n",
              "      <td>72.732000</td>\n",
              "      <td>47.274947</td>\n",
              "      <td>70.191789</td>\n",
              "      <td>47.274947</td>\n",
              "      <td>78.659368</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>65.225739</td>\n",
              "      <td>37.261774</td>\n",
              "      <td>32.023096</td>\n",
              "      <td>37.261774</td>\n",
              "      <td>60.003339</td>\n",
              "      <td>39.127179</td>\n",
              "      <td>72.314713</td>\n",
              "      <td>38.380967</td>\n",
              "      <td>37.618643</td>\n",
              "      <td>38.754115</td>\n",
              "      <td>25.307270</td>\n",
              "      <td>38.007903</td>\n",
              "      <td>56.433809</td>\n",
              "      <td>30.929864</td>\n",
              "      <td>77.910261</td>\n",
              "      <td>31.665725</td>\n",
              "      <td>41.671513</td>\n",
              "      <td>31.049990</td>\n",
              "      <td>20.458017</td>\n",
              "      <td>29.909343</td>\n",
              "      <td>51.885078</td>\n",
              "      <td>54.166539</td>\n",
              "      <td>65.598887</td>\n",
              "      <td>72.703722</td>\n",
              "      <td>37.245496</td>\n",
              "      <td>74.195478</td>\n",
              "      <td>50.303165</td>\n",
              "      <td>70.091687</td>\n",
              "      <td>51.561183</td>\n",
              "      <td>78.268383</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>66.725301</td>\n",
              "      <td>39.621261</td>\n",
              "      <td>32.244810</td>\n",
              "      <td>38.042032</td>\n",
              "      <td>58.565890</td>\n",
              "      <td>39.621261</td>\n",
              "      <td>72.515926</td>\n",
              "      <td>39.884466</td>\n",
              "      <td>36.982380</td>\n",
              "      <td>39.094852</td>\n",
              "      <td>22.506110</td>\n",
              "      <td>38.305237</td>\n",
              "      <td>57.249571</td>\n",
              "      <td>30.672177</td>\n",
              "      <td>77.762945</td>\n",
              "      <td>31.737247</td>\n",
              "      <td>38.035436</td>\n",
              "      <td>30.935382</td>\n",
              "      <td>15.925870</td>\n",
              "      <td>30.672177</td>\n",
              "      <td>43.299534</td>\n",
              "      <td>64.889521</td>\n",
              "      <td>60.671411</td>\n",
              "      <td>77.523239</td>\n",
              "      <td>31.191755</td>\n",
              "      <td>76.997301</td>\n",
              "      <td>44.962748</td>\n",
              "      <td>73.707387</td>\n",
              "      <td>44.227141</td>\n",
              "      <td>86.871166</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "          0          1          2   ...         27         28         29\n",
              "0  66.033564  39.002274  30.227008  ...  72.935459  43.130707  84.485774\n",
              "1  64.332936  34.970077  29.949277  ...  70.266553  45.467915  85.480170\n",
              "2  65.057053  34.909642  30.903789  ...  70.191789  47.274947  78.659368\n",
              "3  65.225739  37.261774  32.023096  ...  70.091687  51.561183  78.268383\n",
              "4  66.725301  39.621261  32.244810  ...  73.707387  44.227141  86.871166\n",
              "\n",
              "[5 rows x 30 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kS7zKDZfQyXe"
      },
      "source": [
        "model = Sequential()\n",
        "\n",
        "# layer set 1\n",
        "model.add(Convolution2D(32, (3,3), padding='same', use_bias=False, input_shape=(96,96,1)))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "\n",
        "model.add(Convolution2D(32, (3,3), padding='same', use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "model.add(MaxPool2D(pool_size=(2, 2)))\n",
        "# result 1 image is converted into 48 X 48 X 32 = 73K \n",
        "\n",
        "# layer set 2\n",
        "model.add(Convolution2D(64, (3,3), padding='same', use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "\n",
        "model.add(Convolution2D(64, (3,3), padding='same', use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "model.add(MaxPool2D(pool_size=(2, 2)))\n",
        "# outputs 24 X 24 X 64 = 36K\n",
        "\n",
        "# layer set 3\n",
        "model.add(Convolution2D(96, (3,3), padding='same', use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "\n",
        "model.add(Convolution2D(96, (3,3), padding='same', use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "model.add(MaxPool2D(pool_size=(2, 2)))\n",
        "# outputs 12 X 12 X 96 = 14K\n",
        "\n",
        "#layer set 4\n",
        "model.add(Convolution2D(128, (3,3),padding='same', use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "\n",
        "model.add(Convolution2D(128, (3,3),padding='same', use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "model.add(MaxPool2D(pool_size=(2, 2)))\n",
        "# outputs 6 X 6 X 128 = 4K\n",
        "\n",
        "#layer set 5\n",
        "model.add(Convolution2D(256, (3,3),padding='same',use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "\n",
        "model.add(Convolution2D(256, (3,3),padding='same',use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "model.add(MaxPool2D(pool_size=(2, 2)))\n",
        "# outputs 3 X 3 X 256 = 2K\n",
        "\n",
        "# different set \n",
        "model.add(Convolution2D(512, (3,3), padding='same', use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "\n",
        "model.add(Convolution2D(512, (3,3), padding='same', use_bias=False))\n",
        "model.add(LeakyReLU(alpha = 0.1))\n",
        "model.add(BatchNormalization())\n",
        "# outputs 3 X 3 X 512 = 4K\n",
        "\n",
        "# Normal layer\n",
        "model.add(Flatten())\n",
        "model.add(Dense(512,activation='relu'))\n",
        "model.add(Dropout(0.1))\n",
        "model.add(Dense(30))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TNMa2rUAVq9u",
        "outputId": "75827bb6-8223-4892-c9a8-de6b2b12310e",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 96, 96, 32)        288       \n",
            "_________________________________________________________________\n",
            "leaky_re_lu (LeakyReLU)      (None, 96, 96, 32)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization (BatchNo (None, 96, 96, 32)        128       \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 96, 96, 32)        9216      \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_1 (LeakyReLU)    (None, 96, 96, 32)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_1 (Batch (None, 96, 96, 32)        128       \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 48, 48, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 48, 48, 64)        18432     \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_2 (LeakyReLU)    (None, 48, 48, 64)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_2 (Batch (None, 48, 48, 64)        256       \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 48, 48, 64)        36864     \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_3 (LeakyReLU)    (None, 48, 48, 64)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_3 (Batch (None, 48, 48, 64)        256       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 24, 24, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_4 (Conv2D)            (None, 24, 24, 96)        55296     \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_4 (LeakyReLU)    (None, 24, 24, 96)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_4 (Batch (None, 24, 24, 96)        384       \n",
            "_________________________________________________________________\n",
            "conv2d_5 (Conv2D)            (None, 24, 24, 96)        82944     \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_5 (LeakyReLU)    (None, 24, 24, 96)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_5 (Batch (None, 24, 24, 96)        384       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 12, 12, 96)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_6 (Conv2D)            (None, 12, 12, 128)       110592    \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_6 (LeakyReLU)    (None, 12, 12, 128)       0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_6 (Batch (None, 12, 12, 128)       512       \n",
            "_________________________________________________________________\n",
            "conv2d_7 (Conv2D)            (None, 12, 12, 128)       147456    \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_7 (LeakyReLU)    (None, 12, 12, 128)       0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_7 (Batch (None, 12, 12, 128)       512       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2 (None, 6, 6, 128)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_8 (Conv2D)            (None, 6, 6, 256)         294912    \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_8 (LeakyReLU)    (None, 6, 6, 256)         0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_8 (Batch (None, 6, 6, 256)         1024      \n",
            "_________________________________________________________________\n",
            "conv2d_9 (Conv2D)            (None, 6, 6, 256)         589824    \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_9 (LeakyReLU)    (None, 6, 6, 256)         0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_9 (Batch (None, 6, 6, 256)         1024      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_4 (MaxPooling2 (None, 3, 3, 256)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_10 (Conv2D)           (None, 3, 3, 512)         1179648   \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_10 (LeakyReLU)   (None, 3, 3, 512)         0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_10 (Batc (None, 3, 3, 512)         2048      \n",
            "_________________________________________________________________\n",
            "conv2d_11 (Conv2D)           (None, 3, 3, 512)         2359296   \n",
            "_________________________________________________________________\n",
            "leaky_re_lu_11 (LeakyReLU)   (None, 3, 3, 512)         0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_11 (Batc (None, 3, 3, 512)         2048      \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 4608)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 512)               2359808   \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 30)                15390     \n",
            "=================================================================\n",
            "Total params: 7,268,670\n",
            "Trainable params: 7,264,318\n",
            "Non-trainable params: 4,352\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "W8E2q0XBWArZ"
      },
      "source": [
        "model.compile(optimizer='adam',\n",
        "              loss='mean_squared_error',\n",
        "              metrics=['mae']\n",
        "              )"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yHuyhXMBWizq",
        "outputId": "ea7bef57-d0e2-44d1-a5dc-ae3d76e1d880",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "model.fit(X_train, y_train, epochs=50, batch_size=256, validation_split=0.2)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            " 2/23 [=>............................] - ETA: 2s - loss: 2302.7283 - mae: 44.5482WARNING:tensorflow:Callbacks method `on_train_batch_end` is slow compared to the batch time (batch time: 0.0933s vs `on_train_batch_end` time: 0.1654s). Check your callbacks.\n",
            "23/23 [==============================] - 7s 301ms/step - loss: 335.2602 - mae: 11.8243 - val_loss: 232.5353 - val_mae: 13.8765\n",
            "Epoch 2/50\n",
            "23/23 [==============================] - 6s 262ms/step - loss: 32.9095 - mae: 4.4633 - val_loss: 27.3105 - val_mae: 4.3038\n",
            "Epoch 3/50\n",
            "23/23 [==============================] - 6s 263ms/step - loss: 26.4669 - mae: 3.9768 - val_loss: 22.0479 - val_mae: 3.7927\n",
            "Epoch 4/50\n",
            "23/23 [==============================] - 6s 264ms/step - loss: 24.3570 - mae: 3.7981 - val_loss: 18.4582 - val_mae: 3.5772\n",
            "Epoch 5/50\n",
            "23/23 [==============================] - 6s 265ms/step - loss: 22.4757 - mae: 3.6452 - val_loss: 28.4977 - val_mae: 4.6107\n",
            "Epoch 6/50\n",
            "23/23 [==============================] - 6s 266ms/step - loss: 20.3168 - mae: 3.4475 - val_loss: 41.1766 - val_mae: 5.6086\n",
            "Epoch 7/50\n",
            "23/23 [==============================] - 6s 267ms/step - loss: 18.7844 - mae: 3.3111 - val_loss: 29.1851 - val_mae: 4.8362\n",
            "Epoch 8/50\n",
            "23/23 [==============================] - 6s 268ms/step - loss: 17.8963 - mae: 3.2183 - val_loss: 22.9636 - val_mae: 3.9895\n",
            "Epoch 9/50\n",
            "23/23 [==============================] - 6s 268ms/step - loss: 17.0765 - mae: 3.1366 - val_loss: 31.4460 - val_mae: 4.8444\n",
            "Epoch 10/50\n",
            "23/23 [==============================] - 6s 269ms/step - loss: 15.8602 - mae: 3.0179 - val_loss: 17.9447 - val_mae: 3.4553\n",
            "Epoch 11/50\n",
            "23/23 [==============================] - 6s 269ms/step - loss: 15.0502 - mae: 2.9347 - val_loss: 28.5967 - val_mae: 4.6276\n",
            "Epoch 12/50\n",
            "23/23 [==============================] - 6s 269ms/step - loss: 15.0976 - mae: 2.9501 - val_loss: 11.5494 - val_mae: 2.6905\n",
            "Epoch 13/50\n",
            "23/23 [==============================] - 6s 270ms/step - loss: 13.3806 - mae: 2.7631 - val_loss: 18.9920 - val_mae: 3.5849\n",
            "Epoch 14/50\n",
            "23/23 [==============================] - 6s 270ms/step - loss: 12.8492 - mae: 2.7132 - val_loss: 19.6198 - val_mae: 3.7408\n",
            "Epoch 15/50\n",
            "23/23 [==============================] - 6s 272ms/step - loss: 12.3757 - mae: 2.6593 - val_loss: 16.5537 - val_mae: 3.3158\n",
            "Epoch 16/50\n",
            "23/23 [==============================] - 6s 272ms/step - loss: 12.5177 - mae: 2.6809 - val_loss: 9.4020 - val_mae: 2.2540\n",
            "Epoch 17/50\n",
            "23/23 [==============================] - 6s 273ms/step - loss: 11.4433 - mae: 2.5572 - val_loss: 5.6332 - val_mae: 1.5765\n",
            "Epoch 18/50\n",
            "23/23 [==============================] - 6s 274ms/step - loss: 12.5592 - mae: 2.6880 - val_loss: 37.2735 - val_mae: 5.2934\n",
            "Epoch 19/50\n",
            "23/23 [==============================] - 6s 274ms/step - loss: 14.8810 - mae: 2.9441 - val_loss: 7.8556 - val_mae: 2.2353\n",
            "Epoch 20/50\n",
            "23/23 [==============================] - 6s 275ms/step - loss: 11.2288 - mae: 2.5369 - val_loss: 4.0673 - val_mae: 1.3183\n",
            "Epoch 21/50\n",
            "23/23 [==============================] - 6s 276ms/step - loss: 10.7408 - mae: 2.4778 - val_loss: 10.5830 - val_mae: 2.4805\n",
            "Epoch 22/50\n",
            "23/23 [==============================] - 6s 276ms/step - loss: 11.5942 - mae: 2.5999 - val_loss: 3.3221 - val_mae: 1.0624\n",
            "Epoch 23/50\n",
            "23/23 [==============================] - 6s 277ms/step - loss: 9.3043 - mae: 2.3087 - val_loss: 3.6176 - val_mae: 1.2461\n",
            "Epoch 24/50\n",
            "23/23 [==============================] - 6s 277ms/step - loss: 9.2687 - mae: 2.3003 - val_loss: 5.0080 - val_mae: 1.3131\n",
            "Epoch 25/50\n",
            "23/23 [==============================] - 6s 277ms/step - loss: 8.7522 - mae: 2.2332 - val_loss: 14.0510 - val_mae: 2.8556\n",
            "Epoch 26/50\n",
            "23/23 [==============================] - 6s 278ms/step - loss: 10.7691 - mae: 2.4696 - val_loss: 5.1636 - val_mae: 1.5080\n",
            "Epoch 27/50\n",
            "23/23 [==============================] - 6s 278ms/step - loss: 9.0525 - mae: 2.2633 - val_loss: 6.3057 - val_mae: 1.6878\n",
            "Epoch 28/50\n",
            "23/23 [==============================] - 6s 279ms/step - loss: 9.5030 - mae: 2.3238 - val_loss: 4.3945 - val_mae: 1.2800\n",
            "Epoch 29/50\n",
            "23/23 [==============================] - 6s 279ms/step - loss: 8.2938 - mae: 2.1673 - val_loss: 3.9085 - val_mae: 1.2583\n",
            "Epoch 30/50\n",
            "23/23 [==============================] - 6s 279ms/step - loss: 9.1041 - mae: 2.2715 - val_loss: 8.3558 - val_mae: 2.2637\n",
            "Epoch 31/50\n",
            "23/23 [==============================] - 6s 280ms/step - loss: 9.4283 - mae: 2.3370 - val_loss: 3.9076 - val_mae: 1.3193\n",
            "Epoch 32/50\n",
            "23/23 [==============================] - 6s 280ms/step - loss: 8.0097 - mae: 2.1402 - val_loss: 3.4161 - val_mae: 1.0226\n",
            "Epoch 33/50\n",
            "23/23 [==============================] - 6s 281ms/step - loss: 7.7819 - mae: 2.1131 - val_loss: 7.8132 - val_mae: 2.2786\n",
            "Epoch 34/50\n",
            "23/23 [==============================] - 6s 282ms/step - loss: 8.1989 - mae: 2.1825 - val_loss: 3.3774 - val_mae: 1.2448\n",
            "Epoch 35/50\n",
            "23/23 [==============================] - 6s 281ms/step - loss: 7.7912 - mae: 2.1316 - val_loss: 3.2254 - val_mae: 1.0884\n",
            "Epoch 36/50\n",
            "23/23 [==============================] - 6s 282ms/step - loss: 8.0083 - mae: 2.1470 - val_loss: 10.6686 - val_mae: 2.4594\n",
            "Epoch 37/50\n",
            "23/23 [==============================] - 6s 282ms/step - loss: 7.3083 - mae: 2.0437 - val_loss: 5.1647 - val_mae: 1.6687\n",
            "Epoch 38/50\n",
            "23/23 [==============================] - 6s 282ms/step - loss: 6.7989 - mae: 1.9695 - val_loss: 75.4731 - val_mae: 8.0296\n",
            "Epoch 39/50\n",
            "23/23 [==============================] - 7s 283ms/step - loss: 20.1318 - mae: 3.3857 - val_loss: 95.9445 - val_mae: 8.6275\n",
            "Epoch 40/50\n",
            "23/23 [==============================] - 7s 283ms/step - loss: 7.9588 - mae: 2.1201 - val_loss: 20.7896 - val_mae: 3.9509\n",
            "Epoch 41/50\n",
            "23/23 [==============================] - 7s 284ms/step - loss: 6.6530 - mae: 1.9505 - val_loss: 8.2630 - val_mae: 2.4265\n",
            "Epoch 42/50\n",
            "23/23 [==============================] - 7s 284ms/step - loss: 6.3310 - mae: 1.9122 - val_loss: 2.6674 - val_mae: 0.9043\n",
            "Epoch 43/50\n",
            "23/23 [==============================] - 7s 284ms/step - loss: 6.8543 - mae: 1.9953 - val_loss: 3.2752 - val_mae: 1.2350\n",
            "Epoch 44/50\n",
            "23/23 [==============================] - 7s 284ms/step - loss: 6.0908 - mae: 1.8734 - val_loss: 3.9905 - val_mae: 1.3362\n",
            "Epoch 45/50\n",
            "23/23 [==============================] - 7s 284ms/step - loss: 6.6030 - mae: 1.9423 - val_loss: 3.7335 - val_mae: 1.2070\n",
            "Epoch 46/50\n",
            "23/23 [==============================] - 7s 284ms/step - loss: 8.6634 - mae: 2.1568 - val_loss: 9.1969 - val_mae: 2.3247\n",
            "Epoch 47/50\n",
            "23/23 [==============================] - 7s 285ms/step - loss: 7.9152 - mae: 2.1199 - val_loss: 3.5318 - val_mae: 1.1507\n",
            "Epoch 48/50\n",
            "23/23 [==============================] - 7s 285ms/step - loss: 6.3746 - mae: 1.8966 - val_loss: 5.2928 - val_mae: 1.5592\n",
            "Epoch 49/50\n",
            "23/23 [==============================] - 7s 286ms/step - loss: 6.0449 - mae: 1.8574 - val_loss: 8.6545 - val_mae: 2.4603\n",
            "Epoch 50/50\n",
            "23/23 [==============================] - 7s 285ms/step - loss: 6.4912 - mae: 1.9257 - val_loss: 3.0630 - val_mae: 1.0075\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7fae6ccc0ef0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BuzioBNCW3-Q"
      },
      "source": [
        "model.save('facialKeyPointsDetection.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uMyFNScjYbNP"
      },
      "source": [
        "#preparing test data\n",
        "timag = []\n",
        "for i in range(0,1783):\n",
        "    timg = test['Image'][i].split(' ')\n",
        "    timg = ['0' if x == '' else x for x in timg]\n",
        "    \n",
        "    timag.append(timg)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XuKb6AzRY2pY"
      },
      "source": [
        "timage_list = np.array(timag,dtype = 'float')\n",
        "X_test = timage_list.reshape(-1,96,96,1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0g-YN7IzY5z2",
        "outputId": "033bb46c-4e99-489a-85b4-dcde2df15e05",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        }
      },
      "source": [
        "plt.imshow(X_test[0].reshape(96,96),cmap = 'gray')\n",
        "plt.show()"
      ],
      "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"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qO8UfE-qY94o"
      },
      "source": [
        "pred = model.predict(X_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "N5ZXdwegZAAm",
        "outputId": "b32e932d-f2c1-4004-e748-f9c1ee002135",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "pred[0]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([65.257225, 36.13281 , 27.524588, 36.964016, 59.330997, 36.96289 ,\n",
              "       72.05108 , 37.86491 , 34.84623 , 37.05759 , 22.176664, 38.450047,\n",
              "       57.559273, 26.537397, 79.54184 , 29.397303, 39.52047 , 25.7828  ,\n",
              "       15.058166, 28.918509, 50.3177  , 51.63561 , 61.569534, 75.13347 ,\n",
              "       33.377007, 75.27023 , 47.984516, 67.72817 , 48.00653 , 85.66994 ],\n",
              "      dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        }
      ]
    }
  ]
}