SVM_FER.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os\n",
    "import argparse\n",
    "import errno\n",
    "import scipy.misc\n",
    "import dlib\n",
    "import cv2\n",
    "from skimage.feature import hog\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.metrics import accuracy_score\n",
    "import sklearn\n",
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "preparing\n"
     ]
    }
   ],
   "source": [
    "print( \"preparing\")\n",
    "predictor = dlib.shape_predictor('shape_predictor_68_face_landmarks.dat')\n",
    "original_labels = [0, 1, 2, 3, 4, 5, 6]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_landmarks(image, rects):\n",
    "    # this function have been copied from http://bit.ly/2cj7Fpq\n",
    "    if len(rects) > 1:\n",
    "        raise BaseException(\"TooManyFaces\")\n",
    "    if len(rects) == 0:\n",
    "        raise BaseException(\"NoFaces\")\n",
    "    return np.matrix([[p.x, p.y] for p in predictor(image, rects[0]).parts()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "SELECTED_LABELS = [0,1,2,3,4,5,6]\n",
    "new_labels = list(set(original_labels) & set(SELECTED_LABELS))\n",
    "nb_images_per_label = list(np.zeros(len(new_labels), 'uint8'))\n",
    "def get_new_label(label, one_hot_encoding=False):\n",
    "    if one_hot_encoding:\n",
    "        new_label = new_labels.index(label)\n",
    "        label = list(np.zeros(len(new_labels), 'uint8'))\n",
    "        label[new_label] = 1\n",
    "        return label\n",
    "    else:\n",
    "        return new_labels.index(label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "importing csv file\n",
      "done\n",
      "Total number of samples:  35887\n"
     ]
    },
    {
     "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>emotion</th>\n",
       "      <th>pixels</th>\n",
       "      <th>Usage</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>70 80 82 72 58 58 60 63 54 58 60 48 89 115 121...</td>\n",
       "      <td>Training</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>151 150 147 155 148 133 111 140 170 174 182 15...</td>\n",
       "      <td>Training</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>231 212 156 164 174 138 161 173 182 200 106 38...</td>\n",
       "      <td>Training</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>24 32 36 30 32 23 19 20 30 41 21 22 32 34 21 1...</td>\n",
       "      <td>Training</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6</td>\n",
       "      <td>4 0 0 0 0 0 0 0 0 0 0 0 3 15 23 28 48 50 58 84...</td>\n",
       "      <td>Training</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   emotion                                             pixels     Usage\n",
       "0        0  70 80 82 72 58 58 60 63 54 58 60 48 89 115 121...  Training\n",
       "1        0  151 150 147 155 148 133 111 140 170 174 182 15...  Training\n",
       "2        2  231 212 156 164 174 138 161 173 182 200 106 38...  Training\n",
       "3        4  24 32 36 30 32 23 19 20 30 41 21 22 32 34 21 1...  Training\n",
       "4        6  4 0 0 0 0 0 0 0 0 0 0 0 3 15 23 28 48 50 58 84...  Training"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print( \"importing csv file\")\n",
    "data = pd.read_csv('fer2013.csv')\n",
    "print(\"done\")\n",
    "print(\"Total number of samples: \", len(data))\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "subset_df = data['pixels'].apply(lambda pixel_str: np.fromstring(pixel_str, sep=' '))\n",
    "subset_df = pd.DataFrame(subset_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "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>pixels</th>\n",
       "      <th>Usage</th>\n",
       "      <th>emotion</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>[70.0, 80.0, 82.0, 72.0, 58.0, 58.0, 60.0, 63....</td>\n",
       "      <td>Training</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>[151.0, 150.0, 147.0, 155.0, 148.0, 133.0, 111...</td>\n",
       "      <td>Training</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>[231.0, 212.0, 156.0, 164.0, 174.0, 138.0, 161...</td>\n",
       "      <td>Training</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>[24.0, 32.0, 36.0, 30.0, 32.0, 23.0, 19.0, 20....</td>\n",
       "      <td>Training</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>[4.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>Training</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              pixels     Usage  emotion\n",
       "0  [70.0, 80.0, 82.0, 72.0, 58.0, 58.0, 60.0, 63....  Training        0\n",
       "1  [151.0, 150.0, 147.0, 155.0, 148.0, 133.0, 111...  Training        0\n",
       "2  [231.0, 212.0, 156.0, 164.0, 174.0, 138.0, 161...  Training        2\n",
       "3  [24.0, 32.0, 36.0, 30.0, 32.0, 23.0, 19.0, 20....  Training        4\n",
       "4  [4.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...  Training        6"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "subset_df = pd.concat([subset_df, data['Usage'], data['emotion']], axis = 1)\n",
    "subset_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "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>...</th>\n",
       "      <th>2294</th>\n",
       "      <th>2295</th>\n",
       "      <th>2296</th>\n",
       "      <th>2297</th>\n",
       "      <th>2298</th>\n",
       "      <th>2299</th>\n",
       "      <th>2300</th>\n",
       "      <th>2301</th>\n",
       "      <th>2302</th>\n",
       "      <th>2303</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>70.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>82.0</td>\n",
       "      <td>72.0</td>\n",
       "      <td>58.0</td>\n",
       "      <td>58.0</td>\n",
       "      <td>60.0</td>\n",
       "      <td>63.0</td>\n",
       "      <td>54.0</td>\n",
       "      <td>58.0</td>\n",
       "      <td>...</td>\n",
       "      <td>159.0</td>\n",
       "      <td>182.0</td>\n",
       "      <td>183.0</td>\n",
       "      <td>136.0</td>\n",
       "      <td>106.0</td>\n",
       "      <td>116.0</td>\n",
       "      <td>95.0</td>\n",
       "      <td>106.0</td>\n",
       "      <td>109.0</td>\n",
       "      <td>82.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>151.0</td>\n",
       "      <td>150.0</td>\n",
       "      <td>147.0</td>\n",
       "      <td>155.0</td>\n",
       "      <td>148.0</td>\n",
       "      <td>133.0</td>\n",
       "      <td>111.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>170.0</td>\n",
       "      <td>174.0</td>\n",
       "      <td>...</td>\n",
       "      <td>105.0</td>\n",
       "      <td>108.0</td>\n",
       "      <td>95.0</td>\n",
       "      <td>108.0</td>\n",
       "      <td>102.0</td>\n",
       "      <td>67.0</td>\n",
       "      <td>171.0</td>\n",
       "      <td>193.0</td>\n",
       "      <td>183.0</td>\n",
       "      <td>184.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>231.0</td>\n",
       "      <td>212.0</td>\n",
       "      <td>156.0</td>\n",
       "      <td>164.0</td>\n",
       "      <td>174.0</td>\n",
       "      <td>138.0</td>\n",
       "      <td>161.0</td>\n",
       "      <td>173.0</td>\n",
       "      <td>182.0</td>\n",
       "      <td>200.0</td>\n",
       "      <td>...</td>\n",
       "      <td>104.0</td>\n",
       "      <td>138.0</td>\n",
       "      <td>152.0</td>\n",
       "      <td>122.0</td>\n",
       "      <td>114.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>88.0</td>\n",
       "      <td>110.0</td>\n",
       "      <td>152.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>24.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>19.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>...</td>\n",
       "      <td>174.0</td>\n",
       "      <td>126.0</td>\n",
       "      <td>132.0</td>\n",
       "      <td>132.0</td>\n",
       "      <td>133.0</td>\n",
       "      <td>136.0</td>\n",
       "      <td>139.0</td>\n",
       "      <td>142.0</td>\n",
       "      <td>143.0</td>\n",
       "      <td>142.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>12.0</td>\n",
       "      <td>34.0</td>\n",
       "      <td>31.0</td>\n",
       "      <td>31.0</td>\n",
       "      <td>31.0</td>\n",
       "      <td>27.0</td>\n",
       "      <td>31.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>30.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 2304 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    0      1      2      3      4      5      6      7      8      9     \\\n",
       "0   70.0   80.0   82.0   72.0   58.0   58.0   60.0   63.0   54.0   58.0   \n",
       "1  151.0  150.0  147.0  155.0  148.0  133.0  111.0  140.0  170.0  174.0   \n",
       "2  231.0  212.0  156.0  164.0  174.0  138.0  161.0  173.0  182.0  200.0   \n",
       "3   24.0   32.0   36.0   30.0   32.0   23.0   19.0   20.0   30.0   41.0   \n",
       "4    4.0    0.0    0.0    0.0    0.0    0.0    0.0    0.0    0.0    0.0   \n",
       "\n",
       "   ...     2294   2295   2296   2297   2298   2299   2300   2301   2302   2303  \n",
       "0  ...    159.0  182.0  183.0  136.0  106.0  116.0   95.0  106.0  109.0   82.0  \n",
       "1  ...    105.0  108.0   95.0  108.0  102.0   67.0  171.0  193.0  183.0  184.0  \n",
       "2  ...    104.0  138.0  152.0  122.0  114.0  101.0   97.0   88.0  110.0  152.0  \n",
       "3  ...    174.0  126.0  132.0  132.0  133.0  136.0  139.0  142.0  143.0  142.0  \n",
       "4  ...     12.0   34.0   31.0   31.0   31.0   27.0   31.0   30.0   29.0   30.0  \n",
       "\n",
       "[5 rows x 2304 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x = np.vstack(subset_df['pixels'].values)\n",
    "x = pd.DataFrame(x)\n",
    "x.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "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>...</th>\n",
       "      <th>2296</th>\n",
       "      <th>2297</th>\n",
       "      <th>2298</th>\n",
       "      <th>2299</th>\n",
       "      <th>2300</th>\n",
       "      <th>2301</th>\n",
       "      <th>2302</th>\n",
       "      <th>2303</th>\n",
       "      <th>Usage</th>\n",
       "      <th>emotion</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>70.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>82.0</td>\n",
       "      <td>72.0</td>\n",
       "      <td>58.0</td>\n",
       "      <td>58.0</td>\n",
       "      <td>60.0</td>\n",
       "      <td>63.0</td>\n",
       "      <td>54.0</td>\n",
       "      <td>58.0</td>\n",
       "      <td>...</td>\n",
       "      <td>183.0</td>\n",
       "      <td>136.0</td>\n",
       "      <td>106.0</td>\n",
       "      <td>116.0</td>\n",
       "      <td>95.0</td>\n",
       "      <td>106.0</td>\n",
       "      <td>109.0</td>\n",
       "      <td>82.0</td>\n",
       "      <td>Training</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>151.0</td>\n",
       "      <td>150.0</td>\n",
       "      <td>147.0</td>\n",
       "      <td>155.0</td>\n",
       "      <td>148.0</td>\n",
       "      <td>133.0</td>\n",
       "      <td>111.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>170.0</td>\n",
       "      <td>174.0</td>\n",
       "      <td>...</td>\n",
       "      <td>95.0</td>\n",
       "      <td>108.0</td>\n",
       "      <td>102.0</td>\n",
       "      <td>67.0</td>\n",
       "      <td>171.0</td>\n",
       "      <td>193.0</td>\n",
       "      <td>183.0</td>\n",
       "      <td>184.0</td>\n",
       "      <td>Training</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>231.0</td>\n",
       "      <td>212.0</td>\n",
       "      <td>156.0</td>\n",
       "      <td>164.0</td>\n",
       "      <td>174.0</td>\n",
       "      <td>138.0</td>\n",
       "      <td>161.0</td>\n",
       "      <td>173.0</td>\n",
       "      <td>182.0</td>\n",
       "      <td>200.0</td>\n",
       "      <td>...</td>\n",
       "      <td>152.0</td>\n",
       "      <td>122.0</td>\n",
       "      <td>114.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>88.0</td>\n",
       "      <td>110.0</td>\n",
       "      <td>152.0</td>\n",
       "      <td>Training</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>24.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>19.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>...</td>\n",
       "      <td>132.0</td>\n",
       "      <td>132.0</td>\n",
       "      <td>133.0</td>\n",
       "      <td>136.0</td>\n",
       "      <td>139.0</td>\n",
       "      <td>142.0</td>\n",
       "      <td>143.0</td>\n",
       "      <td>142.0</td>\n",
       "      <td>Training</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>31.0</td>\n",
       "      <td>31.0</td>\n",
       "      <td>31.0</td>\n",
       "      <td>27.0</td>\n",
       "      <td>31.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>Training</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 2306 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       0      1      2      3      4      5      6      7      8      9  \\\n",
       "0   70.0   80.0   82.0   72.0   58.0   58.0   60.0   63.0   54.0   58.0   \n",
       "1  151.0  150.0  147.0  155.0  148.0  133.0  111.0  140.0  170.0  174.0   \n",
       "2  231.0  212.0  156.0  164.0  174.0  138.0  161.0  173.0  182.0  200.0   \n",
       "3   24.0   32.0   36.0   30.0   32.0   23.0   19.0   20.0   30.0   41.0   \n",
       "4    4.0    0.0    0.0    0.0    0.0    0.0    0.0    0.0    0.0    0.0   \n",
       "\n",
       "    ...      2296   2297   2298   2299   2300   2301   2302   2303     Usage  \\\n",
       "0   ...     183.0  136.0  106.0  116.0   95.0  106.0  109.0   82.0  Training   \n",
       "1   ...      95.0  108.0  102.0   67.0  171.0  193.0  183.0  184.0  Training   \n",
       "2   ...     152.0  122.0  114.0  101.0   97.0   88.0  110.0  152.0  Training   \n",
       "3   ...     132.0  132.0  133.0  136.0  139.0  142.0  143.0  142.0  Training   \n",
       "4   ...      31.0   31.0   31.0   27.0   31.0   30.0   29.0   30.0  Training   \n",
       "\n",
       "   emotion  \n",
       "0        0  \n",
       "1        0  \n",
       "2        2  \n",
       "3        4  \n",
       "4        6  \n",
       "\n",
       "[5 rows x 2306 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x = pd.concat([x, subset_df['Usage'],  subset_df['emotion']], axis = 1)\n",
    "x.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train = x[x[\"Usage\"] == 'Training']\n",
    "x_valid = x[x[\"Usage\"] == 'PublicTest']\n",
    "x_test = x[x[\"Usage\"] == 'PrivateTest']\n",
    "y_train = x_train['emotion']\n",
    "y_valid = x_valid['emotion']\n",
    "y_test = x_test['emotion']\n",
    "del x_train['emotion']\n",
    "del x_valid['emotion']\n",
    "del x_test['emotion']\n",
    "del x_train['Usage']\n",
    "del x_valid['Usage']\n",
    "del x_test['Usage']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "building model...\n",
      "done\n"
     ]
    }
   ],
   "source": [
    "print( \"building model...\")\n",
    "from joblib import load\n",
    "#model = SVC(random_state=42, max_iter=10000, kernel='linear', gamma=0.0001, decision_function_shape = 'ovr')\n",
    "model = load('raw_pixel_svm.joblib')\n",
    "print(\"done\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training on Raw Pixel Data\n",
      "training time = 3000.0 sec\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/svm/base.py:218: ConvergenceWarning: Solver terminated early (max_iter=10000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  % self.max_iter, ConvergenceWarning)\n"
     ]
    }
   ],
   "source": [
    "print(\"Training on Raw Pixel Data\")\n",
    "start_time = time.time()\n",
    "model.fit(x_train, y_train)\n",
    "training_time = time.time() - start_time\n",
    "print( \"training time = {0:.1f} sec\".format(training_time))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  - validation accuracy = 22.0\n"
     ]
    }
   ],
   "source": [
    "validation_accuracy = evaluate(model, x_valid, y_valid)\n",
    "print( \"  - validation accuracy = {0:.1f}\".format(validation_accuracy*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  - test accuracy = 21.6\n"
     ]
    }
   ],
   "source": [
    "test_accuracy = evaluate(model, x_test, y_test)\n",
    "print( \"  - test accuracy = {0:.1f}\".format(test_accuracy*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_learning_curve(estimator, title, X, y, ylim=None, cv=None,\n",
    "                        n_jobs=None, train_sizes=np.linspace(.1, 1.0, 5)):\n",
    "    \"\"\"\n",
    "    Generate a simple plot of the test and training learning curve.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    estimator : object type that implements the \"fit\" and \"predict\" methods\n",
    "        An object of that type which is cloned for each validation.\n",
    "\n",
    "    title : string\n",
    "        Title for the chart.\n",
    "\n",
    "    X : array-like, shape (n_samples, n_features)\n",
    "        Training vector, where n_samples is the number of samples and\n",
    "        n_features is the number of features.\n",
    "\n",
    "    y : array-like, shape (n_samples) or (n_samples, n_features), optional\n",
    "        Target relative to X for classification or regression;\n",
    "        None for unsupervised learning.\n",
    "\n",
    "    ylim : tuple, shape (ymin, ymax), optional\n",
    "        Defines minimum and maximum yvalues plotted.\n",
    "\n",
    "    cv : int, cross-validation generator or an iterable, optional\n",
    "        Determines the cross-validation splitting strategy.\n",
    "        Possible inputs for cv are:\n",
    "          - None, to use the default 3-fold cross-validation,\n",
    "          - integer, to specify the number of folds.\n",
    "          - :term:`CV splitter`,\n",
    "          - An iterable yielding (train, test) splits as arrays of indices.\n",
    "\n",
    "        For integer/None inputs, if ``y`` is binary or multiclass,\n",
    "        :class:`StratifiedKFold` used. If the estimator is not a classifier\n",
    "        or if ``y`` is neither binary nor multiclass, :class:`KFold` is used.\n",
    "\n",
    "        Refer :ref:`User Guide <cross_validation>` for the various\n",
    "        cross-validators that can be used here.\n",
    "\n",
    "    n_jobs : int or None, optional (default=None)\n",
    "        Number of jobs to run in parallel.\n",
    "        ``None`` means 1 unless in a :obj:`joblib.parallel_backend` context.\n",
    "        ``-1`` means using all processors. See :term:`Glossary <n_jobs>`\n",
    "        for more details.\n",
    "\n",
    "    train_sizes : array-like, shape (n_ticks,), dtype float or int\n",
    "        Relative or absolute numbers of training examples that will be used to\n",
    "        generate the learning curve. If the dtype is float, it is regarded as a\n",
    "        fraction of the maximum size of the training set (that is determined\n",
    "        by the selected validation method), i.e. it has to be within (0, 1].\n",
    "        Otherwise it is interpreted as absolute sizes of the training sets.\n",
    "        Note that for classification the number of samples usually have to\n",
    "        be big enough to contain at least one sample from each class.\n",
    "        (default: np.linspace(0.1, 1.0, 5))\n",
    "    \"\"\"\n",
    "    plt.figure()\n",
    "    plt.title(title)\n",
    "    if ylim is not None:\n",
    "        plt.ylim(*ylim)\n",
    "    plt.xlabel(\"Training examples\")\n",
    "    plt.ylabel(\"Score\")\n",
    "    train_sizes, train_scores, test_scores = learning_curve(\n",
    "        estimator, X, y, cv=cv, n_jobs=n_jobs, train_sizes=train_sizes, verbose=3)\n",
    "    train_scores_mean = np.mean(train_scores, axis=1)\n",
    "    train_scores_std = np.std(train_scores, axis=1)\n",
    "    test_scores_mean = np.mean(test_scores, axis=1)\n",
    "    test_scores_std = np.std(test_scores, axis=1)\n",
    "    plt.grid()\n",
    "\n",
    "    plt.fill_between(train_sizes, train_scores_mean - train_scores_std,\n",
    "                     train_scores_mean + train_scores_std, alpha=0.1,\n",
    "                     color=\"r\")\n",
    "    plt.fill_between(train_sizes, test_scores_mean - test_scores_std,\n",
    "                     test_scores_mean + test_scores_std, alpha=0.1, color=\"g\")\n",
    "    plt.plot(train_sizes, train_scores_mean, 'o-', color=\"r\",\n",
    "             label=\"Training score\")\n",
    "    plt.plot(train_sizes, test_scores_mean, 'o-', color=\"g\",\n",
    "             label=\"Cross-validation score\")\n",
    "\n",
    "    plt.legend(loc=\"best\")\n",
    "    return plt\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# https://scikit-learn.org/stable/auto_examples/model_selection/plot_confusion_matrix.html\n",
    "import itertools\n",
    "def plot_confusion_matrix(cm, classes,\n",
    "                          normalize=False,\n",
    "                          title='Confusion matrix',\n",
    "                          cmap=plt.cm.Blues):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    \"\"\"\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "        print(\"Normalized confusion matrix\")\n",
    "    else:\n",
    "        print('Confusion matrix, without normalization')\n",
    "\n",
    "    print(cm)\n",
    "    fig = plt.figure(figsize=(8, 6))\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    fmt = '.2f' if normalize else 'd'\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, format(cm[i, j], fmt),\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    plt.tight_layout()\n",
    "    return fig"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicting\n",
      "Prediction done\n",
      "Constructing the confusion matrix\n"
     ]
    }
   ],
   "source": [
    "print(\"Predicting\")\n",
    "y_predicted_labels = model.predict(x_test)\n",
    "print(\"Prediction done\")\n",
    "print(\"Constructing the confusion matrix\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['raw_pixel_svm.joblib']"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from joblib import dump, load\n",
    "dump(model, 'raw_pixel_svm.joblib') \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Normalized confusion matrix\n",
      "[[0.15 0.03 0.18 0.2  0.21 0.04 0.18]\n",
      " [0.05 0.24 0.09 0.18 0.27 0.05 0.11]\n",
      " [0.1  0.04 0.18 0.28 0.16 0.07 0.17]\n",
      " [0.11 0.03 0.15 0.3  0.2  0.06 0.14]\n",
      " [0.1  0.03 0.21 0.22 0.23 0.06 0.16]\n",
      " [0.09 0.03 0.13 0.33 0.14 0.15 0.13]\n",
      " [0.09 0.03 0.18 0.2  0.23 0.06 0.22]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeYAAAGoCAYAAACTwUbTAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nOUmSJElqxP8DMpHn4fDkHwQAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x14b2b8e10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix = sklearn.metrics.confusion_matrix(y_test, y_predicted_labels)\n",
    "np.set_printoptions(precision=2)\n",
    "labels=[\"Angry\", \"Disgust\", \"Fear\", \"Happy\", \"Sad\", \"Surprise\", \"Neutral\"]\n",
    "# Plot non-normalized confusion matrix\n",
    "fig = plot_confusion_matrix(cnf_matrix, classes=labels,\n",
    "                                     title='Confusion Matrix for Test Data[SVM, Raw Pixel Data]', normalize=True)\n",
    "plt.show()\n",
    "fig.savefig('confusion_matrix_svm_raw_pixel_1.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['Training' 'PublicTest' 'PrivateTest']\n",
      "converting set: Training...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/ipykernel_launcher.py:31: DeprecationWarning: `imsave` is deprecated!\n",
      "`imsave` is deprecated in SciPy 1.0.0, and will be removed in 1.2.0.\n",
      "Use ``imageio.imwrite`` instead.\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/skimage/feature/_hog.py:150: skimage_deprecation: Default value of `block_norm`==`L1` is deprecated and will be changed to `L2-Hys` in v0.15. To supress this message specify explicitly the normalization method.\n",
      "  skimage_deprecation)\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/skimage/feature/_hog.py:248: skimage_deprecation: Argument `visualise` is deprecated and will be changed to `visualize` in v0.16\n",
      "  'be changed to `visualize` in v0.16', skimage_deprecation)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "converting set: PublicTest...\n",
      "converting set: PrivateTest...\n"
     ]
    }
   ],
   "source": [
    "print(data['Usage'].unique())\n",
    "image_height = 48\n",
    "image_width = 48\n",
    "window_size = 24\n",
    "window_step = 6\n",
    "\n",
    "for category in data['Usage'].unique():\n",
    "    print( \"converting set: \" + category + \"...\")\n",
    "    if not os.path.exists(category):\n",
    "        try:\n",
    "            os.makedirs(category)\n",
    "        except OSError as e:\n",
    "            print(\"error!: \", e)\n",
    "    \n",
    "    # get samples and labels of the actual category\n",
    "    category_data = data[data['Usage'] == category]\n",
    "    samples = category_data['pixels'].values\n",
    "    labels = category_data['emotion'].values\n",
    "    \n",
    "    # get images and extract features\n",
    "    images = []\n",
    "    labels_list = []\n",
    "    landmarks = []\n",
    "    hog_features = []\n",
    "    hog_images = []\n",
    "    \n",
    "    for i in range(len(samples)):\n",
    "        try:\n",
    "            image = np.fromstring(samples[i], dtype=int, sep=\" \").reshape((image_height, image_width))\n",
    "            images.append(image)\n",
    "            scipy.misc.imsave('temp.jpg', image)\n",
    "            image2 = cv2.imread('temp.jpg')\n",
    "            \n",
    "            '''Extracting face landmarks here'''\n",
    "            face_rects = [dlib.rectangle(left=1, top=1, right=47, bottom=47)]\n",
    "            face_landmarks = get_landmarks(image2, face_rects)\n",
    "            landmarks.append(face_landmarks) \n",
    "            labels_list.append(get_new_label(labels[i], one_hot_encoding=True))\n",
    "            nb_images_per_label[get_new_label(labels[i])] += 1\n",
    "            \n",
    "            features, hog_image = hog(image, orientations=8, pixels_per_cell=(16, 16),\n",
    "                                            cells_per_block=(1, 1), visualise=True)\n",
    "            hog_features.append(features)\n",
    "            hog_images.append(hog_image)\n",
    "        except Exception as e:\n",
    "            print( \"error in image: \" + str(i) + \" - \" + str(e))\n",
    "    \n",
    "    np.save(\"./\"+category+'/images.npy', images)\n",
    "    np.save(\"./\"+category+'/landmarks.npy', landmarks)\n",
    "    np.save(\"./\"+category+'/labels.npy', labels_list)\n",
    "    np.save(\"./\"+category+'/hog_features.npy', hog_features)\n",
    "    np.save(\"./\"+category+'/hog_images.npy', hog_images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_dict = dict()\n",
    "validation_dict = dict()\n",
    "test_dict = dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_dict['X'] = np.load('./Training' + '/landmarks.npy')\n",
    "data_dict['X'] = np.array([x.flatten() for x in data_dict['X']])\n",
    "data_dict['X'] = np.concatenate((data_dict['X'], np.load('./Training' + '/hog_features.npy')), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of training samples:  28709\n"
     ]
    }
   ],
   "source": [
    "print(\"Total number of training samples: \", len(data_dict['X']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_dict['Y'] = np.load('./Training' +  '/labels.npy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "validation_dict['X'] = np.load('./PublicTest' + '/landmarks.npy')\n",
    "validation_dict['X'] = np.array([x.flatten() for x in validation_dict['X']])\n",
    "validation_dict['Y'] = np.load('./PublicTest'  + '/labels.npy')\n",
    "validation_dict['X'] = np.concatenate((validation_dict['X'], np.load('./PublicTest' + '/hog_features.npy')), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of training samples:  3589\n"
     ]
    }
   ],
   "source": [
    "print(\"Total number of training samples: \", len(validation_dict['X']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_dict['X'] = np.load('./PrivateTest' + '/landmarks.npy')\n",
    "test_dict['X'] = np.array([x.flatten() for x in test_dict['X']])\n",
    "test_dict['Y'] = np.load('./PrivateTest'  + '/labels.npy')\n",
    "test_dict['X'] = np.concatenate((test_dict['X'], np.load('./PrivateTest' + '/hog_features.npy')), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of samples:  3589\n"
     ]
    }
   ],
   "source": [
    "print(\"Number of samples: \", len(test_dict['X']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "building model...\n",
      "done\n"
     ]
    }
   ],
   "source": [
    "print( \"building model...\")\n",
    "model = SVC(random_state=42, max_iter=10000, kernel='rbf', gamma=0.0001, decision_function_shape = 'ovr')\n",
    "print(\"done\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Getting targets\n",
      "Done\n"
     ]
    }
   ],
   "source": [
    "print(\"Getting targets\")\n",
    "dict_y=np.ndarray(len(data_dict['Y']))\n",
    "count=0\n",
    "for row in range(len(data_dict['Y'])):\n",
    "    temp = data_dict['Y'][row][0]\n",
    "    dict_y[count] = temp\n",
    "    count+=1\n",
    "print(\"Done\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Training - with Face Landmarks and HOG Features\")\n",
    "start_time = time.time()\n",
    "model.fit(data_dict['X'], dict_y)\n",
    "training_time = time.time() - start_time\n",
    "print( \"training time = {0:.1f} sec\".format(training_time))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate(model, X, Y):\n",
    "    predicted_Y = model.predict(X)\n",
    "    accuracy = accuracy_score(Y, predicted_Y)\n",
    "    return accuracy\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "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",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     0\n",
       "0  1.0\n",
       "1  1.0\n",
       "2  0.0\n",
       "3  0.0\n",
       "4  0.0"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_df = pd.DataFrame(dict_y)\n",
    "y_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "dict_v=np.ndarray(len(validation_dict['Y']))\n",
    "count=0\n",
    "for row in range(len(validation_dict['Y'])):\n",
    "    temp = validation_dict['Y'][row][0]\n",
    "    dict_v[count] = temp\n",
    "    count+=1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  - validation accuracy = 87.0\n"
     ]
    }
   ],
   "source": [
    "validation_accuracy = evaluate(model, validation_dict['X'], dict_v)\n",
    "print( \"  - validation accuracy = {0:.1f}\".format(validation_accuracy*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "dict_t=np.ndarray(len(test_dict['Y']))\n",
    "count=0\n",
    "for row in range(len(test_dict['Y'])):\n",
    "    temp = test_dict['Y'][row][0]\n",
    "    dict_t[count] = temp\n",
    "    count+=1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  - test accuracy = 86.3\n"
     ]
    }
   ],
   "source": [
    "test_accuracy = evaluate(model, test_dict['X'], dict_t)\n",
    "print( \"  - test accuracy = {0:.1f}\".format(test_accuracy*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "'''Using ONLY HOG features'''\n",
    "data_hog_train = np.load('./Training' + '/hog_features.npy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training time = 34.6 sec\n"
     ]
    }
   ],
   "source": [
    "start_time = time.time()\n",
    "model.fit(data_hog_train, dict_y)\n",
    "training_time = time.time() - start_time\n",
    "print( \"training time = {0:.1f} sec\".format(training_time))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_predicted_labels = model.predict(data_test_hog)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Normalized confusion matrix\n",
      "[[1. 0. 0. 0. 0. 0. 0.]\n",
      " [1. 0. 0. 0. 0. 0. 0.]\n",
      " [1. 0. 0. 0. 0. 0. 0.]\n",
      " [1. 0. 0. 0. 0. 0. 0.]\n",
      " [1. 0. 0. 0. 0. 0. 0.]\n",
      " [1. 0. 0. 0. 0. 0. 0.]\n",
      " [1. 0. 0. 0. 0. 0. 0.]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAd8AAAGoCAYAAAAHJ+8hAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12fe4b6a0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix = sklearn.metrics.confusion_matrix(y_test, y_predicted_labels)\n",
    "np.set_printoptions(precision=2)\n",
    "labels=[\"Angry\", \"Disgust\", \"Fear\", \"Happy\", \"Sad\", \"Surprise\", \"Neutral\"]\n",
    "# Plot non-normalized confusion matrix\n",
    "fig = plot_confusion_matrix(cnf_matrix, classes=labels,\n",
    "                                     title='Confusion Matrix for Test Data[SVM, HOG Features]', normalize=True)\n",
    "plt.show()\n",
    "fig.savefig('confusion_matrix_svm_HOG.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  - validation accuracy = 87.0\n"
     ]
    }
   ],
   "source": [
    "data_validation_hog = np.load('./PublicTest' + '/hog_features.npy')\n",
    "validation_accuracy = evaluate(model, data_validation_hog, dict_v)\n",
    "print( \"  - validation accuracy = {0:.1f}\".format(validation_accuracy*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_test_hog = np.load('./PrivateTest' + '/hog_features.npy')\n",
    "test_accuracy = evaluate(model, data_test_hog, dict_t)\n",
    "print( \"  - test accuracy = {0:.1f}\".format(test_accuracy*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "'''Only Face Landmarks'''\n",
    "\n",
    "data_dict['X'] = np.load('./Training' + '/landmarks.npy')\n",
    "data_dict['X'] = np.array([x.flatten() for x in data_dict['X']])\n",
    "\n",
    "validation_dict['X'] = np.load('./PublicTest' + '/landmarks.npy')\n",
    "validation_dict['X'] = np.array([x.flatten() for x in validation_dict['X']])\n",
    "\n",
    "test_dict['X'] = np.load('./PrivateTest' + '/landmarks.npy')\n",
    "test_dict['X'] = np.array([x.flatten() for x in test_dict['X']])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = SVC(random_state=42, max_iter=10000, kernel='linear', gamma=0.0001, decision_function_shape = 'ovr')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_dict['Y'] = np.load('./Training' +  '/labels.npy')\n",
    "dict_y = np.ndarray(len(data_dict['Y']))\n",
    "count=0\n",
    "for row in range(len(data_dict['Y'])):\n",
    "    temp = data_dict['Y'][row][0]\n",
    "    dict_y[count] = temp\n",
    "    count+=1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training time = 117.1 sec\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/svm/base.py:218: ConvergenceWarning: Solver terminated early (max_iter=10000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  % self.max_iter, ConvergenceWarning)\n"
     ]
    }
   ],
   "source": [
    "start = time.time()\n",
    "model.fit(data_dict['X'], dict_y)\n",
    "training_time = time.time() - start_time\n",
    "print( \"training time = {0:.1f} sec\".format(training_time))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_dict['Y'] = np.load('./PrivateTest'  + '/labels.npy')\n",
    "dict_t=np.ndarray(len(test_dict['Y']))\n",
    "count=0\n",
    "for row in range(len(test_dict['Y'])):\n",
    "    temp = test_dict['Y'][row][0]\n",
    "    dict_t[count] = temp\n",
    "    count+=1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Prediction begins\n",
      "Prediction complete\n"
     ]
    }
   ],
   "source": [
    "print(\"Prediction begins\")\n",
    "y_predicted_labels = model.predict(test_dict['X'])\n",
    "print(\"Prediction complete\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Normalized confusion matrix\n",
      "[[0.42 0.58 0.   0.   0.   0.   0.  ]\n",
      " [0.38 0.62 0.   0.   0.   0.   0.  ]\n",
      " [0.43 0.57 0.   0.   0.   0.   0.  ]\n",
      " [0.44 0.56 0.   0.   0.   0.   0.  ]\n",
      " [0.49 0.51 0.   0.   0.   0.   0.  ]\n",
      " [0.44 0.56 0.   0.   0.   0.   0.  ]\n",
      " [0.48 0.52 0.   0.   0.   0.   0.  ]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAd8AAAGoCAYAAAAHJ+8hAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nZp58nXPOuTT7f4D+vGx9IcWxAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x200e9ef60>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix = sklearn.metrics.confusion_matrix(y_test, y_predicted_labels)\n",
    "np.set_printoptions(precision=2)\n",
    "labels=[\"Angry\", \"Disgust\", \"Fear\", \"Happy\", \"Sad\", \"Surprise\", \"Neutral\"]\n",
    "# Plot non-normalized confusion matrix\n",
    "fig = plot_confusion_matrix(cnf_matrix, classes=labels,\n",
    "                                     title='Confusion Matrix for Test Data[SVM, Face Landmarks]', normalize=True)\n",
    "plt.show()\n",
    "fig.savefig('confusion_matrix_svm_FACE_LANDMARKS.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  - test accuracy = 47.1%\n",
      "training accuracy:  = 48.0%\n"
     ]
    }
   ],
   "source": [
    "test_accuracy = evaluate(model, test_dict['X'], dict_t)\n",
    "print(\"  - test accuracy = {0:.1f}%\".format(test_accuracy*100))\n",
    "print(\"  - training accuracy:  = {0:.1f}%\".format(model.score(data_dict['X'], dict_y)*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  - validation accuracy = 48.0%\n"
     ]
    }
   ],
   "source": [
    "validation_dict['Y'] = np.load('./PrivateTest'  + '/labels.npy')\n",
    "dict_t=np.ndarray(len(validation_dict['Y']))\n",
    "count=0\n",
    "for row in range(len(validation_dict['Y'])):\n",
    "    temp = validation_dict['Y'][row][0]\n",
    "    dict_t[count] = temp\n",
    "    count+=1\n",
    "validation_accuracy = evaluate(model, validation_dict['X'], dict_v)\n",
    "print(\"  - validation accuracy = {0:.1f}%\".format(validation_accuracy*100))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}