Automatic VoTT Bounding Box suggestions with Google Cloud AutoML Vision model

normalize-csv.ipynb

{
 "metadata": {
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5-final"
  },
  "orig_nbformat": 2,
  "kernelspec": {
   "name": "Python 3.8.5 64-bit ('.venv')",
   "display_name": "Python 3.8.5 64-bit ('.venv')",
   "metadata": {
    "interpreter": {
     "hash": "d275ea291513c142d309cf1729fea5f027504164ee1a26fc0a6595b29a650687"
    }
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2,
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "                        image        xmin        ymin        xmax        ymax  \\\n",
       "0         -0-0uBnUSnc.mp4#t=0  378.079149  219.691963  393.018451  229.739478   \n",
       "1  -0-0uBnUSnc.mp4#t=0.333333  289.442322  247.933643  308.510313  267.861507   \n",
       "2  -0-0uBnUSnc.mp4#t=0.333333  320.733384  248.592176  443.941940  272.260692   \n",
       "3  -0-0uBnUSnc.mp4#t=0.333333  286.215996  249.400038  479.883723  280.570265   \n",
       "4  -0-0uBnUSnc.mp4#t=0.666667  294.331551  245.000853  313.399542  264.928717   \n",
       "\n",
       "    label  \n",
       "0  nozzle  \n",
       "1  nozzle  \n",
       "2   print  \n",
       "3    raft  \n",
       "4  nozzle  "
      ],
      "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>image</th>\n      <th>xmin</th>\n      <th>ymin</th>\n      <th>xmax</th>\n      <th>ymax</th>\n      <th>label</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>-0-0uBnUSnc.mp4#t=0</td>\n      <td>378.079149</td>\n      <td>219.691963</td>\n      <td>393.018451</td>\n      <td>229.739478</td>\n      <td>nozzle</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>-0-0uBnUSnc.mp4#t=0.333333</td>\n      <td>289.442322</td>\n      <td>247.933643</td>\n      <td>308.510313</td>\n      <td>267.861507</td>\n      <td>nozzle</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>-0-0uBnUSnc.mp4#t=0.333333</td>\n      <td>320.733384</td>\n      <td>248.592176</td>\n      <td>443.941940</td>\n      <td>272.260692</td>\n      <td>print</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>-0-0uBnUSnc.mp4#t=0.333333</td>\n      <td>286.215996</td>\n      <td>249.400038</td>\n      <td>479.883723</td>\n      <td>280.570265</td>\n      <td>raft</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>-0-0uBnUSnc.mp4#t=0.666667</td>\n      <td>294.331551</td>\n      <td>245.000853</td>\n      <td>313.399542</td>\n      <td>264.928717</td>\n      <td>nozzle</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
     },
     "metadata": {},
     "execution_count": 2
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "# load VoTT CSV export\n",
    "# notice: coordinates are absolute\n",
    "df = pd.read_csv('/home/leigh/datasets/spaghetti/labeled/vott-csv-export/spaghetti_v1-export.csv')\n",
    "df.head()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 8926 entries, 0 to 8925\nData columns (total 6 columns):\n #   Column  Non-Null Count  Dtype  \n---  ------  --------------  -----  \n 0   image   8926 non-null   object \n 1   xmin    8926 non-null   float64\n 2   ymin    8926 non-null   float64\n 3   xmax    8926 non-null   float64\n 4   ymax    8926 non-null   float64\n 5   label   8926 non-null   object \ndtypes: float64(4), object(2)\nmemory usage: 418.5+ KB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "finished 0 / 8926\n",
      "finished 1000 / 8926\n",
      "...\n"
     ]
    }
   ],
   "source": [
    "\n",
    "import cv2\n",
    "\n",
    "base_path = '/home/leigh/datasets/spaghetti/labeled/vott-csv-export/'\n",
    "\n",
    "LOG_INTERVAL=2000\n",
    "\n",
    "# convert absolute coordinates to relative coordinates in [0, 1] range\n",
    "for index, row in df.iterrows():\n",
    "    if index % LOG_INTERVAL == 0:\n",
    "        print(f'finished {index} / {len(df)}')\n",
    "    filename = row['image_path'].split('/')[-1]\n",
    "    img = cv2.imread(f'{base_path}{filename}')\n",
    "    height, width, channels = img.shape\n",
    "    df.at[index, 'x1_n'] = row['x1'] / width\n",
    "    df.at[index, 'x2_n']= row['x2'] / width  \n",
    "    df.at[index, 'y1_n'] = row['y1'] / height\n",
    "    df.at[index, 'y2_n'] = row['y2'] / height\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# write CSV with columns expected by AutoML Vision\n",
    "df['none'] = ''\n",
    "df.to_csv('/home/leigh/datasets/spaghetti/labeled/vott-csv-export/spaghetti_v1-normalized-export.csv', \n",
    "    columns=['set', 'image_path', 'label', 'x1_n', 'y1_n', 'none', 'none', 'x2_n', 'y2_n', 'none', 'none'],\n",
    "    index=False\n",
    "    )\n"
   ]
  }
 ]
}