evaluate.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import os\n",
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "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>Unnamed: 0</th>\n",
       "      <th>filename</th>\n",
       "      <th>id</th>\n",
       "      <th>head_x</th>\n",
       "      <th>head_y</th>\n",
       "      <th>eye_r_x</th>\n",
       "      <th>eye_r_y</th>\n",
       "      <th>eye_l_x</th>\n",
       "      <th>eye_l_y</th>\n",
       "      <th>neck_x</th>\n",
       "      <th>...</th>\n",
       "      <th>right_foot_y</th>\n",
       "      <th>left_hip_x</th>\n",
       "      <th>left_hip_y</th>\n",
       "      <th>left_knee_x</th>\n",
       "      <th>left_knee_y</th>\n",
       "      <th>left_ankle_x</th>\n",
       "      <th>left_ankle_y</th>\n",
       "      <th>left_foot_x</th>\n",
       "      <th>left_foot_y</th>\n",
       "      <th>file_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>534600936</td>\n",
       "      <td>85.615750</td>\n",
       "      <td>24.769218</td>\n",
       "      <td>77.062070</td>\n",
       "      <td>61.667427</td>\n",
       "      <td>88.425110</td>\n",
       "      <td>62.263695</td>\n",
       "      <td>86.871704</td>\n",
       "      <td>...</td>\n",
       "      <td>180.75030</td>\n",
       "      <td>106.741930</td>\n",
       "      <td>154.17407</td>\n",
       "      <td>116.441864</td>\n",
       "      <td>166.45120</td>\n",
       "      <td>116.197310</td>\n",
       "      <td>192.77790</td>\n",
       "      <td>122.117540</td>\n",
       "      <td>194.96344</td>\n",
       "      <td>000000.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>534600946</td>\n",
       "      <td>87.690560</td>\n",
       "      <td>24.574066</td>\n",
       "      <td>82.641840</td>\n",
       "      <td>61.296080</td>\n",
       "      <td>94.084390</td>\n",
       "      <td>61.998566</td>\n",
       "      <td>90.680660</td>\n",
       "      <td>...</td>\n",
       "      <td>223.82635</td>\n",
       "      <td>106.381890</td>\n",
       "      <td>157.01001</td>\n",
       "      <td>113.821236</td>\n",
       "      <td>171.32074</td>\n",
       "      <td>110.395454</td>\n",
       "      <td>201.43866</td>\n",
       "      <td>115.376854</td>\n",
       "      <td>208.65576</td>\n",
       "      <td>000001.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>534600943</td>\n",
       "      <td>87.526825</td>\n",
       "      <td>25.332672</td>\n",
       "      <td>81.923400</td>\n",
       "      <td>61.537690</td>\n",
       "      <td>93.457855</td>\n",
       "      <td>62.006410</td>\n",
       "      <td>90.272220</td>\n",
       "      <td>...</td>\n",
       "      <td>239.30927</td>\n",
       "      <td>105.748535</td>\n",
       "      <td>158.89398</td>\n",
       "      <td>112.388336</td>\n",
       "      <td>176.69437</td>\n",
       "      <td>109.467150</td>\n",
       "      <td>211.18924</td>\n",
       "      <td>110.902740</td>\n",
       "      <td>220.45410</td>\n",
       "      <td>000002.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>534600952</td>\n",
       "      <td>87.487650</td>\n",
       "      <td>23.486847</td>\n",
       "      <td>85.206410</td>\n",
       "      <td>61.637527</td>\n",
       "      <td>96.168690</td>\n",
       "      <td>61.579163</td>\n",
       "      <td>91.132030</td>\n",
       "      <td>...</td>\n",
       "      <td>237.48965</td>\n",
       "      <td>106.520210</td>\n",
       "      <td>151.71670</td>\n",
       "      <td>115.251240</td>\n",
       "      <td>162.20319</td>\n",
       "      <td>98.887380</td>\n",
       "      <td>192.78906</td>\n",
       "      <td>97.089270</td>\n",
       "      <td>200.74863</td>\n",
       "      <td>000003.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>534600949</td>\n",
       "      <td>87.804050</td>\n",
       "      <td>22.924591</td>\n",
       "      <td>84.868774</td>\n",
       "      <td>60.159150</td>\n",
       "      <td>95.819520</td>\n",
       "      <td>60.853760</td>\n",
       "      <td>90.475555</td>\n",
       "      <td>...</td>\n",
       "      <td>220.97986</td>\n",
       "      <td>106.368164</td>\n",
       "      <td>157.46863</td>\n",
       "      <td>110.289630</td>\n",
       "      <td>178.60913</td>\n",
       "      <td>107.886610</td>\n",
       "      <td>208.90955</td>\n",
       "      <td>111.741940</td>\n",
       "      <td>216.75750</td>\n",
       "      <td>000004.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4650</th>\n",
       "      <td>4650</td>\n",
       "      <td>NaN</td>\n",
       "      <td>473472707</td>\n",
       "      <td>137.458020</td>\n",
       "      <td>40.468803</td>\n",
       "      <td>113.374210</td>\n",
       "      <td>52.571983</td>\n",
       "      <td>95.676940</td>\n",
       "      <td>44.350320</td>\n",
       "      <td>94.365845</td>\n",
       "      <td>...</td>\n",
       "      <td>248.12036</td>\n",
       "      <td>111.646530</td>\n",
       "      <td>141.75159</td>\n",
       "      <td>131.974660</td>\n",
       "      <td>168.06177</td>\n",
       "      <td>127.665115</td>\n",
       "      <td>246.08704</td>\n",
       "      <td>129.139570</td>\n",
       "      <td>246.45554</td>\n",
       "      <td>004650.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4651</th>\n",
       "      <td>4651</td>\n",
       "      <td>NaN</td>\n",
       "      <td>473460564</td>\n",
       "      <td>144.630680</td>\n",
       "      <td>41.528595</td>\n",
       "      <td>97.812540</td>\n",
       "      <td>45.876846</td>\n",
       "      <td>93.479805</td>\n",
       "      <td>45.813430</td>\n",
       "      <td>100.280130</td>\n",
       "      <td>...</td>\n",
       "      <td>247.06421</td>\n",
       "      <td>101.240290</td>\n",
       "      <td>182.40555</td>\n",
       "      <td>135.293800</td>\n",
       "      <td>167.42252</td>\n",
       "      <td>127.279670</td>\n",
       "      <td>247.43420</td>\n",
       "      <td>128.733030</td>\n",
       "      <td>247.47760</td>\n",
       "      <td>004651.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4652</th>\n",
       "      <td>4652</td>\n",
       "      <td>NaN</td>\n",
       "      <td>473472709</td>\n",
       "      <td>136.277250</td>\n",
       "      <td>41.421013</td>\n",
       "      <td>104.795990</td>\n",
       "      <td>47.400490</td>\n",
       "      <td>99.439620</td>\n",
       "      <td>45.730340</td>\n",
       "      <td>96.868650</td>\n",
       "      <td>...</td>\n",
       "      <td>245.32181</td>\n",
       "      <td>114.196380</td>\n",
       "      <td>126.63453</td>\n",
       "      <td>131.877370</td>\n",
       "      <td>168.72714</td>\n",
       "      <td>126.293430</td>\n",
       "      <td>248.32239</td>\n",
       "      <td>128.374560</td>\n",
       "      <td>248.54700</td>\n",
       "      <td>004652.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4653</th>\n",
       "      <td>4653</td>\n",
       "      <td>NaN</td>\n",
       "      <td>473472740</td>\n",
       "      <td>144.335370</td>\n",
       "      <td>41.354294</td>\n",
       "      <td>92.647736</td>\n",
       "      <td>46.069534</td>\n",
       "      <td>91.244220</td>\n",
       "      <td>47.141130</td>\n",
       "      <td>98.848850</td>\n",
       "      <td>...</td>\n",
       "      <td>250.74724</td>\n",
       "      <td>120.467740</td>\n",
       "      <td>149.04265</td>\n",
       "      <td>135.144040</td>\n",
       "      <td>155.86345</td>\n",
       "      <td>128.544460</td>\n",
       "      <td>250.70119</td>\n",
       "      <td>131.068630</td>\n",
       "      <td>250.48060</td>\n",
       "      <td>004653.png</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4654</th>\n",
       "      <td>4654</td>\n",
       "      <td>NaN</td>\n",
       "      <td>473460572</td>\n",
       "      <td>144.944090</td>\n",
       "      <td>41.119263</td>\n",
       "      <td>85.524060</td>\n",
       "      <td>44.083847</td>\n",
       "      <td>86.151184</td>\n",
       "      <td>44.654312</td>\n",
       "      <td>97.631960</td>\n",
       "      <td>...</td>\n",
       "      <td>249.86847</td>\n",
       "      <td>116.384980</td>\n",
       "      <td>144.53873</td>\n",
       "      <td>133.868730</td>\n",
       "      <td>161.46716</td>\n",
       "      <td>128.107600</td>\n",
       "      <td>244.94312</td>\n",
       "      <td>129.648200</td>\n",
       "      <td>244.99960</td>\n",
       "      <td>004654.png</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4655 rows × 46 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Unnamed: 0  filename         id      head_x     head_y     eye_r_x  \\\n",
       "0              0       NaN  534600936   85.615750  24.769218   77.062070   \n",
       "1              1       NaN  534600946   87.690560  24.574066   82.641840   \n",
       "2              2       NaN  534600943   87.526825  25.332672   81.923400   \n",
       "3              3       NaN  534600952   87.487650  23.486847   85.206410   \n",
       "4              4       NaN  534600949   87.804050  22.924591   84.868774   \n",
       "...          ...       ...        ...         ...        ...         ...   \n",
       "4650        4650       NaN  473472707  137.458020  40.468803  113.374210   \n",
       "4651        4651       NaN  473460564  144.630680  41.528595   97.812540   \n",
       "4652        4652       NaN  473472709  136.277250  41.421013  104.795990   \n",
       "4653        4653       NaN  473472740  144.335370  41.354294   92.647736   \n",
       "4654        4654       NaN  473460572  144.944090  41.119263   85.524060   \n",
       "\n",
       "        eye_r_y    eye_l_x    eye_l_y      neck_x  ...  right_foot_y  \\\n",
       "0     61.667427  88.425110  62.263695   86.871704  ...     180.75030   \n",
       "1     61.296080  94.084390  61.998566   90.680660  ...     223.82635   \n",
       "2     61.537690  93.457855  62.006410   90.272220  ...     239.30927   \n",
       "3     61.637527  96.168690  61.579163   91.132030  ...     237.48965   \n",
       "4     60.159150  95.819520  60.853760   90.475555  ...     220.97986   \n",
       "...         ...        ...        ...         ...  ...           ...   \n",
       "4650  52.571983  95.676940  44.350320   94.365845  ...     248.12036   \n",
       "4651  45.876846  93.479805  45.813430  100.280130  ...     247.06421   \n",
       "4652  47.400490  99.439620  45.730340   96.868650  ...     245.32181   \n",
       "4653  46.069534  91.244220  47.141130   98.848850  ...     250.74724   \n",
       "4654  44.083847  86.151184  44.654312   97.631960  ...     249.86847   \n",
       "\n",
       "      left_hip_x  left_hip_y  left_knee_x  left_knee_y  left_ankle_x  \\\n",
       "0     106.741930   154.17407   116.441864    166.45120    116.197310   \n",
       "1     106.381890   157.01001   113.821236    171.32074    110.395454   \n",
       "2     105.748535   158.89398   112.388336    176.69437    109.467150   \n",
       "3     106.520210   151.71670   115.251240    162.20319     98.887380   \n",
       "4     106.368164   157.46863   110.289630    178.60913    107.886610   \n",
       "...          ...         ...          ...          ...           ...   \n",
       "4650  111.646530   141.75159   131.974660    168.06177    127.665115   \n",
       "4651  101.240290   182.40555   135.293800    167.42252    127.279670   \n",
       "4652  114.196380   126.63453   131.877370    168.72714    126.293430   \n",
       "4653  120.467740   149.04265   135.144040    155.86345    128.544460   \n",
       "4654  116.384980   144.53873   133.868730    161.46716    128.107600   \n",
       "\n",
       "      left_ankle_y  left_foot_x  left_foot_y   file_name  \n",
       "0        192.77790   122.117540    194.96344  000000.png  \n",
       "1        201.43866   115.376854    208.65576  000001.png  \n",
       "2        211.18924   110.902740    220.45410  000002.png  \n",
       "3        192.78906    97.089270    200.74863  000003.png  \n",
       "4        208.90955   111.741940    216.75750  000004.png  \n",
       "...            ...          ...          ...         ...  \n",
       "4650     246.08704   129.139570    246.45554  004650.png  \n",
       "4651     247.43420   128.733030    247.47760  004651.png  \n",
       "4652     248.32239   128.374560    248.54700  004652.png  \n",
       "4653     250.70119   131.068630    250.48060  004653.png  \n",
       "4654     244.94312   129.648200    244.99960  004654.png  \n",
       "\n",
       "[4655 rows x 46 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# img = '000000.png' \n",
    "# img_id = 534600936\n",
    "# head = [85.56447, 24.732826]\n",
    "# eye_r = [77.036545, 61.643562]\n",
    "# eye_l = [88.15402, 62.289528]\n",
    "# neck = [87.04971, 83.33402] \n",
    "\n",
    "# img_id = []\n",
    "# head = [] \n",
    "\n",
    "# f = open('C:/Users/user/Desktop/models/vitpose/20221015/result_keypoints.txt', 'r')\n",
    "#     # print(f)\n",
    "# while True:\n",
    "#     line = f.readline()\n",
    "#     if not line: break \n",
    "#     img_id_axis = line.split('//')[1].replace(' ', '')\n",
    "#     img_id.append(int(img_id_axis))\n",
    "#     head_axis = line.split('//')[2].split('],')[0].replace('[', '').replace(' ', '').split(',')\n",
    "#     head.append([head_axis[0], head_axis[1]])\n",
    "    \n",
    "# f.close() \n",
    "\n",
    "# head, img_id\n",
    "# img_id\n",
    "df = pd.read_csv(os.path.join('/Users/sehwan/Dropbox/test', 'result_keypoints.csv'))\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calc_distance(arr1, arr2):\n",
    "    arr1 = np.array(arr1, dtype=float)\n",
    "    arr2 = np.array(arr2, dtype=float)\n",
    "    \n",
    "    return np.linalg.norm(arr1 - arr2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# path = 'E:/22.9.21/GM/openpose/custom_dataset/annotations'\n",
    "# paths = ['train', 'test', 'valid']\n",
    "# paths = os.path.join(path, 'test_baby_keypoints.json')\n",
    "paths = '/Users/sehwan/Dropbox/test/test_baby_keypoints.json'\n",
    "\n",
    "\n",
    "# coco = [] \n",
    "# for p in paths:\n",
    "#     valid_path = os.path.join(path, f'{p}_baby_keypoints.json')\n",
    "    \n",
    "with open(paths, 'r') as f:\n",
    "    json_data = json.load(f)\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.01 threhold : 5.177228786251343\n",
      "0.02 threhold : 14.52201933404941\n",
      "0.03 threhold : 24.146079484425346\n",
      "0.04 threhold : 33.77013963480129\n",
      "0.05 threhold : 41.997851772287866\n",
      "0.06 threhold : 49.23737916219119\n",
      "0.07 threhold : 55.85392051557465\n",
      "0.08 threhold : 61.568206229860365\n",
      "0.09 threhold : 65.77873254564985\n",
      "0.1 threhold : 69.49516648764768\n",
      "0.11 threhold : 72.73899033297529\n",
      "0.12 threhold : 76.00429645542427\n",
      "0.13 threhold : 78.34586466165413\n",
      "0.14 threhold : 80.90225563909775\n",
      "0.15 threhold : 83.05048335123523\n",
      "0.16 threhold : 85.06981740064447\n",
      "0.17 threhold : 86.61654135338345\n",
      "0.18 threhold : 87.9484425349087\n",
      "0.19 threhold : 89.30182599355531\n",
      "0.2 threhold : 90.37593984962406\n",
      "0.21 threhold : 91.1922663802363\n",
      "0.22 threhold : 91.94414607948443\n",
      "0.23 threhold : 92.43823845327604\n",
      "0.24 threhold : 92.69602577873255\n",
      "0.25 threhold : 92.9967776584318\n",
      "0.26 threhold : 93.27604726100967\n",
      "0.27 threhold : 93.72717508055854\n",
      "0.28 threhold : 94.07089151450053\n",
      "0.29 threhold : 94.30719656283566\n",
      "0.3 threhold : 94.39312567132117\n",
      "0.31 threhold : 94.6938775510204\n",
      "0.32 threhold : 94.84425349087003\n",
      "0.33 threhold : 94.95166487647691\n",
      "0.34 threhold : 95.03759398496241\n",
      "0.35 threhold : 95.03759398496241\n",
      "0.36 threhold : 95.10204081632652\n",
      "0.37 threhold : 95.1235230934479\n",
      "0.38 threhold : 95.16648764769066\n",
      "0.39 threhold : 95.2094522019334\n",
      "0.4 threhold : 95.23093447905478\n",
      "0.41 threhold : 95.23093447905478\n",
      "0.42 threhold : 95.23093447905478\n",
      "0.43 threhold : 95.25241675617615\n",
      "0.44 threhold : 95.25241675617615\n",
      "0.45 threhold : 95.25241675617615\n",
      "0.46 threhold : 95.25241675617615\n",
      "0.47 threhold : 95.25241675617615\n",
      "0.48 threhold : 95.25241675617615\n",
      "0.49 threhold : 95.25241675617615\n",
      "0.5 threhold : 95.25241675617615\n",
      "0.51 threhold : 95.25241675617615\n",
      "0.52 threhold : 95.25241675617615\n",
      "0.53 threhold : 95.25241675617615\n",
      "0.54 threhold : 95.25241675617615\n",
      "0.55 threhold : 95.25241675617615\n",
      "0.56 threhold : 95.25241675617615\n",
      "0.57 threhold : 95.25241675617615\n",
      "0.58 threhold : 95.25241675617615\n",
      "0.59 threhold : 95.25241675617615\n",
      "0.6 threhold : 95.25241675617615\n",
      "0.61 threhold : 95.25241675617615\n",
      "0.62 threhold : 95.25241675617615\n",
      "0.63 threhold : 95.25241675617615\n",
      "0.64 threhold : 95.25241675617615\n",
      "0.65 threhold : 95.25241675617615\n",
      "0.66 threhold : 95.25241675617615\n",
      "0.67 threhold : 95.25241675617615\n",
      "0.68 threhold : 95.25241675617615\n",
      "0.69 threhold : 95.25241675617615\n",
      "0.7 threhold : 95.25241675617615\n",
      "0.71 threhold : 95.25241675617615\n",
      "0.72 threhold : 95.25241675617615\n",
      "0.73 threhold : 95.25241675617615\n",
      "0.74 threhold : 95.25241675617615\n",
      "0.75 threhold : 95.25241675617615\n",
      "0.76 threhold : 95.25241675617615\n",
      "0.77 threhold : 95.25241675617615\n",
      "0.78 threhold : 95.25241675617615\n",
      "0.79 threhold : 95.25241675617615\n",
      "0.8 threhold : 95.25241675617615\n",
      "0.81 threhold : 95.25241675617615\n",
      "0.82 threhold : 95.25241675617615\n",
      "0.83 threhold : 95.27389903329752\n",
      "0.84 threhold : 95.27389903329752\n",
      "0.85 threhold : 95.27389903329752\n",
      "0.86 threhold : 95.27389903329752\n",
      "0.87 threhold : 95.31686358754028\n",
      "0.88 threhold : 95.3813104189044\n",
      "0.89 threhold : 95.44575725026853\n",
      "0.9 threhold : 95.48872180451127\n",
      "0.91 threhold : 95.61761546723953\n",
      "0.92 threhold : 95.68206229860365\n",
      "0.93 threhold : 95.72502685284641\n",
      "0.94 threhold : 95.74650912996778\n",
      "0.95 threhold : 95.83243823845328\n",
      "0.96 threhold : 95.87540279269604\n",
      "0.97 threhold : 95.8968850698174\n",
      "0.98 threhold : 95.96133190118152\n",
      "0.99 threhold : 96.02577873254565\n",
      "1.0 threhold : 96.04726100966703\n"
     ]
    }
   ],
   "source": [
    "pchk_list = []\n",
    "# idx = - \n",
    "for ii in range(100):     \n",
    "    ii = (ii + 1) / 100\n",
    "    pckh = 0 \n",
    "    total = 0 \n",
    "    for ann in json_data['annotations']:\n",
    "        idx = img_id.index(ann['image_id'])\n",
    "        \n",
    "        thr = calc_distance([ann['keypoints'][0], ann['keypoints'][1]], [ann['keypoints'][9], ann['keypoints'][10]]) * ii\n",
    "        # print(thr / 2)\n",
    "        dist = calc_distance(head[idx], [ann['keypoints'][0], ann['keypoints'][1]])\n",
    "        \n",
    "        if dist <= thr:\n",
    "            pckh += 1\n",
    "            \n",
    "        total += 1\n",
    "        \n",
    "    \n",
    "        \n",
    "        # coco_data = {}\n",
    "        # /print(f\"------------------->{ann['image_id']}\")\n",
    "        # print(f\"'------------------->{ann['bbox']}\")\n",
    "        \n",
    "        # coco_data['bbox'] = ann['bbox']\n",
    "        # coco_data['category_id'] = ann['category_id']\n",
    "        # coco_data['image_id'] = ann['image_id']\n",
    "        # coco_data['score'] = 1\n",
    "        \n",
    "        # coco.append(coco_data)\n",
    "        # if img_id == ann['image_id']:\n",
    "        # print(f\"'------------------->{ann['keypoints']}\")\n",
    "        # print(f\"head : {calc_distance(head, [ann['keypoints'][0], ann['keypoints'][1]])}\")\n",
    "        # print(f\"eye_r : {calc_distance(eye_r, [ann['keypoints'][3], ann['keypoints'][4]])}\")\n",
    "        # print(f\"eye_l : {calc_distance(eye_l, [ann['keypoints'][6], ann['keypoints'][7]])}\")\n",
    "        # print(f\"neck : {calc_distance(neck, [ann['keypoints'][9], ann['keypoints'][10]])}\")\n",
    "    pchk_list.append(pckh / total * 100)            \n",
    "    print(f'{ii} threhold : {pckh / total * 100}')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "100"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(pchk_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'N')"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1440x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "# pchk_list\n",
    "plt.figure(figsize=(20, 5))\n",
    "plt.plot(range(100), pchk_list)\n",
    "plt.ylabel('PCKh@n')\n",
    "plt.xlabel('N')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "path = 'C:/Users/user/Desktop/models/vitpose/20221015'\n",
    "logs = os.path.join(path, '20221014_104238.log.json')\n",
    "\n",
    "data = [] \n",
    "# with open(logs, 'r') as f:\n",
    "#     data.append(json.load(f))\n",
    "    \n",
    "# data    \n",
    "for j in open(logs, 'r'):\n",
    "    data.append(json.loads(j))\n",
    "    \n",
    "# data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "# data =[] \n",
    "# data[1]\n",
    "\n",
    "# {'mode': 'train', 'epoch': 1, 'iter': 850, 'lr': 1e-05, 'memory': 4046, 'data_time': 0.00042, 'heatmap_loss': 0.00148, 'acc_pose': 0.7301, 'loss': 0.00148, 'grad_norm': 0.00872, 'time': 0.10553}\n",
    "\n",
    "# {\"mode\": \"val\", \"epoch\": 60, \"iter\": 609, \"lr\": 1e-05, \"AP\": 0.9742, \"AP .5\": 1.0, \"AP .75\": 0.98993, \"AP (M)\": -1.0, \"AP (L)\": 0.9742, \"AR\": 0.97719, \"AR .5\": 1.0, \"AR .75\": 0.99384, \"AR (M)\": -1.0, \"AR (L)\": 0.97719}\n",
    "\n",
    "\n",
    "\n",
    "heatmap_loss = []\n",
    "acc_pose = [] \n",
    "loss = [] \n",
    "grad_norm = []\n",
    "\n",
    "\n",
    "ap = []\n",
    "ap_5 = []\n",
    "ap_75 = [] \n",
    "ap_m = [] \n",
    "ap_l = [] \n",
    "\n",
    "ar = [] \n",
    "ar_5 = [] \n",
    "ar_75 = [] \n",
    "ar_m = [] \n",
    "ar_l = [] \n",
    "\n",
    "for ii in range(1, len(data)-1):\n",
    "    # print(data[ii])\n",
    "    # ii_data = data[ii]\n",
    "    # print(ii_data[ii_data['mode'] == 'train']['mode'])\n",
    "    if data[ii]['mode'] == 'train':\n",
    "        # print(1)\n",
    "        if data[ii]['iter'] >= 850:\n",
    "            # print(data[ii]['epoch'], ' Epoch')\n",
    "            heatmap_loss.append(data[ii]['heatmap_loss'])\n",
    "            acc_pose.append(data[ii]['acc_pose'])\n",
    "            loss.append(data[ii]['loss'])\n",
    "            grad_norm.append(data[ii]['grad_norm'])\n",
    "            \n",
    "    elif data[ii]['mode'] == 'val':\n",
    "        ap.append(data[ii]['AP'])\n",
    "        ap_5.append(data[ii]['AP .5'])\n",
    "        ap_75.append(data[ii]['AP .75']) \n",
    "        ap_m.append(data[ii]['AP (M)']) \n",
    "        ap_l.append(data[ii]['AP (L)']) \n",
    "\n",
    "        ar.append(data[ii]['AR']) \n",
    "        ar_5.append(data[ii]['AR .5']) \n",
    "        ar_75.append(data[ii]['AR .75']) \n",
    "        ar_m.append(data[ii]['AR (M)']) \n",
    "        ar_l.append(data[ii]['AR (L)']) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# heatmap_loss = []\n",
    "# acc_pose = [] \n",
    "# loss = [] \n",
    "# grad_norm = [] \n",
    "\n",
    "plt.plot(heatmap_loss, label='heatmap_loss')\n",
    "plt.plot(acc_pose, label='acc_pose')\n",
    "plt.plot(loss, '--', label='loss')\n",
    "plt.plot(grad_norm, '--', label='grad_norm')\n",
    "plt.legend(loc=(1.05, 0.02))\n",
    "plt.title('Train Evaluation')\n",
    "plt.xlabel('Epoch')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x_axis = range(0, len(ap) * 10, 10)\n",
    "plt.plot(x_axis, ap, label='AP')\n",
    "plt.plot(x_axis, ap_5, label='AP .5')\n",
    "plt.plot(x_axis, ap_75, label='AP .75')\n",
    "plt.plot(x_axis, ap_m, label='AP (M)')\n",
    "plt.plot(x_axis, ap_l, label='AP (L)')\n",
    "\n",
    "plt.plot(x_axis, ap, '--', label='AR')\n",
    "plt.plot(x_axis, ap_5, '--', label='AR .5')\n",
    "plt.plot(x_axis, ap_75, '--', label='AR .75')\n",
    "plt.plot(x_axis, ap_m, '--', label='AR (M)')\n",
    "plt.plot(x_axis, ap_l, '--', label='AR (L)')\n",
    "\n",
    "\n",
    "plt.legend(loc=(1.05, 0.02))\n",
    "plt.title('Validation Evaluation')\n",
    "plt.xlabel('Epoch')\n",
    "# plt.ylabel('Epoch')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "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": "3.9.12"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "17cf7ee649b5372d63375ed9a5f807db8792a6d95aaf1648689b5d9f3b845e3c"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}