draw_bbox.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import h5py\n",
    "from PIL import Image\n",
    "%pylab inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_attrs(digit_struct_mat_file, index):\n",
    "    \"\"\"\n",
    "    Returns a dictionary which contains keys: label, left, top, width and height, each key has multiple values.\n",
    "    \"\"\"\n",
    "    attrs = {}\n",
    "    f = digit_struct_mat_file\n",
    "    item = f['digitStruct']['bbox'][index].item()\n",
    "    for key in ['label', 'left', 'top', 'width', 'height']:\n",
    "        attr = f[item][key]\n",
    "        values = [f[attr[i].item()][0][0]\n",
    "                  for i in range(len(attr))] if len(attr) > 1 else [attr[0][0]]\n",
    "        attrs[key] = values\n",
    "    return attrs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "path_to_dir = 'data/test'\n",
    "path_to_digit_struct_mat_file = os.path.join(path_to_dir, 'digitStruct.mat')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "index 2: data/test/3.png\n"
     ]
    }
   ],
   "source": [
    "path_to_image_file = os.path.join(path_to_dir, '3.png')\n",
    "index = int(path_to_image_file.split('/')[-1].split('.')[0]) - 1\n",
    "print('index %d: %s' % (index, path_to_image_file))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "min_left=61, min_top=6, max_right=72, max_bottom=22\n",
      "center_x=66.5, center_y=14.0, max_side=16\n",
      "bbox: left=58.5, top=6.0, width=16, height=16\n",
      "cropped: left=56, top=4, width=21, height=21\n"
     ]
    }
   ],
   "source": [
    "with h5py.File(path_to_digit_struct_mat_file, 'r') as digit_struct_mat_file:\n",
    "    attrs = get_attrs(digit_struct_mat_file, index)\n",
    "    length = len(attrs['label'])\n",
    "    attrs_left, attrs_top, attrs_width, attrs_height = map(lambda x: [int(i) for i in x],\n",
    "                                                       [attrs['left'], attrs['top'], attrs['width'], attrs['height']])\n",
    "    min_left, min_top, max_right, max_bottom = (min(attrs_left),\n",
    "                                                min(attrs_top),\n",
    "                                                max(map(lambda x, y: x + y, attrs_left, attrs_width)),\n",
    "                                                max(map(lambda x, y: x + y, attrs_top, attrs_height)))\n",
    "    center_x, center_y, max_side = ((min_left + max_right) / 2.0,\n",
    "                                    (min_top + max_bottom) / 2.0,\n",
    "                                    max(max_right - min_left, max_bottom - min_top))\n",
    "    bbox_left, bbox_top, bbox_width, bbox_height = (center_x - max_side / 2.0, \n",
    "                                                    center_y - max_side / 2.0, \n",
    "                                                    max_side,\n",
    "                                                    max_side)\n",
    "    cropped_left, cropped_top, cropped_width, cropped_height = (int(round(bbox_left - 0.15 * bbox_width)),\n",
    "                                                                int(round(bbox_top - 0.15 * bbox_height)),\n",
    "                                                                int(round(bbox_width * 1.3)),\n",
    "                                                                int(round(bbox_height * 1.3)))\n",
    "print('min_left=%d, min_top=%d, max_right=%d, max_bottom=%d' % (min_left, min_top, max_right, max_bottom))\n",
    "print('center_x=%.1f, center_y=%.1f, max_side=%d' % (center_x, center_y, max_side))\n",
    "print('bbox: left=%.1f, top=%.1f, width=%d, height=%d' % (bbox_left, bbox_top, bbox_width, bbox_height))\n",
    "print('cropped: left=%d, top=%d, width=%d, height=%d' % (cropped_left, cropped_top, cropped_width, cropped_height))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAACJCAYAAAAmJCOuAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJztnXuQHdWd37+/7nvnLQlhQAjhsrB5lePF4GjBju01YLAxq0WA2cReV5atcpn9I654492KSaUqhqokRaq8JNny1pblR8wmXjuJDUZQXmQti18VAhYYWB7mYRsHkEAGgYQe87i3T/64d6a/v1/36ek7M5p7mfl9qlTqvn369OnTp3v6fPv3kBACHMdxnJVH0u8GOI7jOMcGf8A7juOsUPwB7ziOs0LxB7zjOM4KxR/wjuM4KxR/wDuO46xQ/AHvOI6zQlnUA15ELhORJ0XkGRG5fqka5TiO4yweWaijk4ikAJ4CcCmA5wH8FMDHQwiPL13zHMdxnIXSWMS+5wN4JoTwSwAQkW8B2AYg+oAfGR4KE2NjnRURtU0koeX8d/v3h1cT2qfRbKpyjUZ+akmal0uSfNn+aRPUQ6iBXEfWzlS5VquFsoJJoo/E58h1ZJmuL2R5weo/zFK6WDjhept0zRLrpYV6RNfs9Uixutes0L7oeRw7eLx2CCVLS8QCzm8hPVIYD9ETqXeG1aO6qoXle4baZ6X3V8eS2ndHzXJ82Hp173nxhZdDCCf2Wv1iHvCbADxH688DuKBqh4mxMWy96AMAAGmkatvQ0MjccpLm27JMd0Crna+PjIzNLZ+w4WRV7sST8vWxNXm5kbHRueV2aKl9JNVtmvs90xeuSX9M+Bl86NAhVe6Vfb/JV0Jex8jQkCrHzTh0IK/j6OFJVW5menpueWoyX5ZUX0ZBfh5B/bHkPxD6nPhPSdUfjyH6w8k3dyZZWXEAQFIxhrOISihmp9gfFpF4Od5m28d/6IXawOeeVNywofCwzskifT4+MqzrCNy+8t8BQLLyDmwj3ueNBr/w1FNi43+84+WGzNgLoV26T7BtDeXHqhp7HdEgRnlf2HFed/+Urm+oGsAEjxc+bFLR/e2Z+DVk/t1//NyvaxUstOkYIyLXichuEdk9OTU9/w6O4zjOkrCYN/gXALyZ1k/t/qYIIWwHsB0A3rR+fci6fxkb5q+xfnuILQNA5A3B/OXXb6v5snrDqzmlMi/waNN0TiraOjKczxbSNH+jGm5oOYlfepKQX5KhhnmDn5kpXV63br2uL+M3ifwNgd9m2ubNJtZfwbz5NlM9+5ij8g0+/h6h37JpH9FDM6PrnvJMyxxXjyPaVvGm32xEzqnq/afinPSB8jY0GnofNWa5HwrSXHnVdd/gi2++kf2q3pDpPLgr7Rt8Xep++9P3btUb/EKOW+/tmU+4atbK4zw6Dm17yh9nnW1VG2uymDf4nwI4Q0ROE5EhAB8DsGPRLXIcx3GWhAW/wYcQWiLyaQA7AaQAvhZCeGzJWuY4juMsisVINAghfA/A95aoLY7jOM4SsqgHfK+ELMPk0SkAgIj+4BomSDdms5IK3W2GrAvShj6V8fHxueWRsXHaUq6ndX+IbNNKVoM0W5bhGqYNzeHcMihr5XoaWwwBRnMl2W1MtVvXATK1HB+dUOWi6mbKGqHpV1qXiOYIaNNPifYXkCjNm8xUK6TXFovNBTPamBWNKZdEvrdYUzc+x5glj+kjdZ0qNHilG1N72u0ZXS5q+WFtg8tN9qQR/4aUSDO6LaZ/B7YminzrMk1Q1kiAvr6xb19VbaiivkVMnLraOBtJqVI1vzUpq7SKvqzaVPsbQQUeqsBxHGeF4g94x3GcFcqySjRZCJic7kxT7dQOh8lxh6fW1olHbcvnN62Wns60ySO03c7LNSq8DuJmWBXT8QrnEGVuSHVYs0t1TjTtFtPWZspTQDLzs45TSUzO4Lm1PlfeR3n7mql0OhxxBjMSSKIsAOP9l4LNzvLfJ6enTPvK95fCubKcUc8bMebQlBVkHZK42nF5Kkt4LNJ1SoxsQvW3lcmkcbpBeZ9bBxzlYW0HWS34nOLSKJ9uq2U0Bh5HFc5baSgfY23rUUoHSyrPqVzOqDqPyurUMOK665nOirqe8WdYpZ3kEuBv8I7jOCsUf8A7juOsUJZXoskCJo92pBgr0WRtXieri1RPYRKSFiTNrRLsVJFlmZDVm2LxjLdq+qa82TgAmJkOqthjaspmLD/Iw7ER8mm8tcpRbU1omzmlaJgUnhkaGYANFFr0ad9KIMrQJerRqz1PE/DU1TaJZCc1q417OlfFTFHjir0vTaeocpFYL1V+kxzAzsoKLKmwN2Kz4PVJXq4RWQ0wMoUao+Ya0hibmaFrWAgIVu5xWdXHda2YYu0p1BcxEKnoBoRW3AuV74e63q+6TWYMqIbUjNHDshPtklnZiVdJki3EIKoKYlMTf4N3HMdZoSzrG7zjrBYu+/QfYvzll/rdjIHj8IknYecXb+l3M1YN/oB3nGPA+Msv4Tvf2mkkGjthpkBwb1CJxgZGi7Vndv+r/9lHouWdpWd5H/AhN2e0GnwI5OVHAlba1E1skGbVJk0uM9ZGykxMBUXn41aYZFV42iVUX1VccG0mySaKxusz4fjtpMuac+c+4+iUNl547PsBFwuiO0x3V/ymbTbLvXgzcwGUHzCVsyaJfO6cCEUlS0GZOWT398KDiPeJRfcz3piFB285bBKrziPykEtTAd9iie3XtFyztXBb+YGaZbqP2NNTRyLUlafKozn/vcHnZ+7P2MN/uhVveDvwNxB62Kdxfdkq5uoeasa9y3WUWP7DEo/g2QyUd8J+HEr4fNk8uSJefcT0s+AwH2IrtkL3ZHUcx3Ei+APecRxnhbK8wcaQTyOztpFHotMq8zcoxLbZKRtP/ctPszC5jJhXLTAveTSgUkHPJGE1oeOmTX1OnOiioTxPtYckTyNVWjgyXbSp8jLSgzNOVwYrvZQnAymYXZKNKMsZklSYP9J1So1s0o7pwwVHVpL3qIqiCRpP6WOYMcUrGUsO5d60kiZgGTAzmTtESXjUVlNdRia/7ZlclrEyFpsGa4nFXGuSeZp0PbI0fs/Ecibb1Ip8f6YRU8MkSaLezYXrxN65tW/E3r8lJAUzSaWjlCyVHbX8uJUx0li6qogxt1D8Dd5xHGeF4lY0jtMHPvXzP8JvZvZVljl4/UH8+b1/jht/eCOe/PST2Po3W7FmeA22b92OLV/egi986AvY8/oe3HzvzXjhsy9gy/YtOPNNZ+KGC2/ARbdchC9t/RJ279mNLz/4ZRy8/iBOufkUXLj5Qlz3rutwxbeuwDeu/gZuf+pOfPPRbyJ8PkBuFHz8HR/H1jO34hO3fgI7PrYD2x/cjh88+wPs+ewerL1pbWV7T2puwJfP/h9L2U3OIpGFxGVeKGvG14R3nX0ugOJXehVjnWWKhrWiycuNr10zt/yW0zarcps3b6ZyeVz14RH+Iq7lB5ZABHGJpjGUSyKc2/TQoUOq3Kuvvooy1q1bp9abTTouHWx4WAcR0xINSx12LseyDAdyIonGWH5wX1jPTEZNcTmgm80OT9saHDM/1RJNyvlQSWpqmSk8WysFFa/eBDmLxICvmqqn0Yl3xQQ34v06u9+2qy/F7bfuAptCZiEPqHf1P/webjvnjrl15flork1rOrcwm5rKg7BNT+mAbJynly297LmzhzQvs6et9aJO0xSb37YZz/7iWS2lDQ3PLW975MP47m/dhTJm23DF1R/Gjlt3lpYBqmWYhQRQs1Y0psZ8qVB1VrEtRjxIX4wqIxpuw9XXfOSBEMKWui0ptshxnIHitM1v6XcTFOuPP67fTXB6xB/wjjOgrF+/bv5Cy8jPfvpQv5vg9Ig/4B1nQHnwZ4/0uwmK91/8vn43wemR5TWTDAHtdkx04qiC7FFqTbxIU6YkH60ZrafzOpuTDYO0ftsU9kRDuZkUYAP2R363VVd4imqTRdL+jI4XM3m0rqsxPb0dtFkdM5Pl+q02s7TRPOk6kSdxe0bnGwVtY3l+KNXfFVifTxp0XGv6qfqvKnkHt533Md8I2JOVbgNtlleRjKFCmJ31Eu5ch/y4qfn+wJ9RMnWtO/v8zvvfhx/9+CfK/HF6OtfxJyePqPqmpvJrkHJE1sL3rvzAzWbez0HlGjZjrxnw6MOPodVqqWilVRFP1f5sElqls1f0eagZ1VFlA65remhMWFW5mp8pVS7puu/Oqu7Fe65a/A3ecQaUx5/4eb+boHj94Ov9boLTI/6Ad5wBxVpl9ZtLL7+k301wemT5g411JRprZdZolpsYFYKIsZUee/i1jUTT5mQgNHUKw1TKBiLKl9VUvdKbtiKa3gI80XimWMhPGUlc0jQer0JBlCAcxC2vr2WmpCwXBB2VTJV77cBreftYBpvW/c/JGVKqrmkkmmYjvx6NRi4XTBy3XpVjL0GW6Qr5UOkcC3l/uX2RQHBKPqgM9lSV6GL2/7ZOOmISfiQs2XAQse5U/+KLPoC/3XmXqiMjGWDayJJsJjkyNpEfx/QDewkPUT7f4WZ+LVi6AYBGOoTv33E3hodGux66s+doZURKtBMLeod21DywKphXbYtu8qDNjORT2yi8hnRSlJq4n8kkOe64XtmgqqB/dfE3eMcZUP52Z7lNeb9421lv7XcTnB6Z9wEvIl8TkX0i8ij9dryI7BKRp7v/r6+qw3Gc3jn99NP73QTFyOjw/IWcgaKORPN1AF8E8Nf02/UA7g4h3CQi13fXPzdvTSJAd5qamfkbGcQgozmNtZhgq4sW7TQ9qb36WLLhaej0dD7FbTTM1JXWOQ59aiw6WN5gOWNkZESVGxsbm1vm6fPhyaOq3MRoXm6EPAMTiQfm4mk3n1+nffk5Zkq6otyvZto+RdYZr+z/zdzyi/u0O/3MZH4er7yyP/99ysRvJxlrJM3PaWJ0QpVbO5G7v284+ZS55faBA6rcuvW5kw1bViVGokkiceMLOYA5h29NKU3F+K4IaDXridopn+8zOTmpyrFXakp5DmYlxeGhIbRaLaXF79+f93m7pe8NHgcNWl4zPq7KsTXP9HTeBvagHWrqsTwyMoJfPfVrjI2NqTjvkzPTqIO1ouH+40eBze3LcAz5Ks9TbdpWETAuun+FpY/K82u2RYMLmnVergh6tyzBxkIIPwKw3/y8DcBs3q1bAFy5+KY4jsM8/Mhg2cFf8P6ePeWdPrNQDX5DCGFvd/lFABtiBUXkOhHZLSK7Z1ozsWKO4xgu+eBgWa08/siT/W6C0yOL/sga7Dy0uH17CGFLCGFLs9GMFXMcx/DAA7v73QRFuxV3lHMGk4WaSb4kIhtDCHtFZCOA6rinXQLyPI1ikxAENm0q/322jtK6KxIFsFmS9myzf984Q0TvkeHqambx6IU6kqYlieQYnTG2pCGUe8Mqc0Bjp8rfCFgrPnpUfy94+eVcrTuwn00mbTIL1uBH55aPDOtvJUeP5Boum6OetPFkVY5NXVlDDrBmeuXXLclsfltajw4qax5bXtBGfxT+n8eeqU55slIds96qR44exfT0NA4fPjy37eDBg3k5891DfYtR37R0+4bI+5Sve4NMWJtD+joljSbec/45+OGu/6NPouoeAuvV+psFX6dqU0Fa4YibS6BPV5k467ypNc0VI+UKb9HqnNgce+lZ6Bv8DgDXdpevBXD70jTHcZxZLvrAhf1ugqL4cHcGnTpmkt8EcC+As0TkeRH5JICbAFwqIk8DuKS77jjOErLjzjvmL7SMnHG228G/0ZhXogkhfDyy6YO9Hy7QtFlPSHgaGWheVvSUk5rllm7CU5gOxhw9F3hIO32lLfqw6pzyv81sLtppU7lnJssZmZFoOHnE4UO5LHPggI4/cmD/q7QtlwtC2wS0ItPSjDa1p3Rb25OUu5Vkk/G1OlTuxASZ45lEKAx3H48pG+hLrbcX5zFozSRr5+KN7DMrR537znfioYcfVuaULJnNGO9hPt+pkXwfTpIDAO203Mu72Si2oZxyj02LMiutOHe1T0WxpZFlyo9V8KyNyC11r+Gg4J6sjuPU4umf/7LfTXB6xB/wjjOgPPTww/1uguIDl/6TfjfB6ZFlDzYWk2hUMeX1VrWNl800OWJFoONXL338Zaau9BL7O1ucDpafe6ulp8naEIfrYFlHW0lMUyzxw4fzOOOHjEQzeSS3sJkmCxgRbQKbkIlIO6F45sYXgoOSHX49t7aZPKy9PlmGSpvl1hmADqSlJAIjSbEXczsWTKr01zpk5v+5BqlVtjRqkEfjbL7dK664Ajt27FByEnuYNhJ93cfJY3V4mAOHWY/ofD1r1bsHMgHuv/chZAKkKUmjU3FpVFmEqQBbyYK8NG1sd92+3t9Vq/Khck5mc6T4PjVlqOXE3+AdZ0C55557+t0EBScKcd4Y+APecQYUG7K335xz3tv73QSnR/wB7zgDygUXXNDvJih+/IP7+t0Ep0eWNycrcj28GIgtkkuzYP5Iy1m5yWSxvnJzShu5bhA1NCZqdmbsx4TPl3OZqg8TJpon57cl8zsbJfIoaeNTR3IdPzGCZtrM65+mfm4bs7oW6cuNNNf72zP6uOp6UhXBCLOxa2i9OZcLbk+rVaXqcw7Vzm25c+cuJElDRTPlt/p0WIfvnZjII3WOU8KPQrRRutZsgsmmkdZMstVq4R3nnI1HH9FpBAv9PeD3kKZe4p7l4lg8f/wN3nEGlPPOO7ffTVBMHp2cv5AzUPgD3nEGFBsHqN888/Sz/W6C0yPLayaJ3HzRTlbre7LGzCQrPF6VMkFT4cLfNyoo5eaFS0FiTLDEBrWK7phfrhBJ/tFZ58Bt+XJKvd4OJvgWtUmZ7wXT1owCfbXjAaNaZJaYJFm0HHdFRjlGrQlbzETO9t0Q5T1l00gr0XBCmGPpf1jl3RjbNvv7E0905BCde5g9f00yC1pnM8k01R9rZ9qTtEw5Xmn5yJR+W29OTWHbVR/G7bftVO1JG9aruFz2UFKh2HJMlVRSdZ9E9rP3VjRwWMWzoC6x+7giWJkyx6zpTdsL/gbvOAPKVVdt63cTFDvv+mG/m+D0iD/gHWdAueuu7/e7CYq1aybmL+QMFH3zZLW5M5V3HU+nzbRHW86gZrnIVNh8tLaWOHWoCqgUPa5Zj3u8VkEySqPi7zQHHmuzBZGWXhqSD4Uhigs+1NCWGsPkSTndoABgbV2fknzoWqTmuvNxmySvaI/j3LsT0HJN0ZO1fBxx3HNAyx5W6jhW2OEQm57Pjt01E2tx5PBkLAy98hYGgJT6j2UZe348LpVU1S7/Hej05T/6rbPw4q59yGh/6/q02Dyide+hBcPPiYVIIIvdH+Ycj3F8Mn+Dd5wB5Zxz3tHvJiju3vXjfjfB6RF/wDvOgLJr19/3uwmK8999Xr+b4PTIslvRzFIZv12lXYsH5qqyookfuObfNFXO7lMuy1hLjYTkB+G5mJnaBbXOafmqGsjWFNaSgS2SuG6yvDHnlCqpJK9vuDGiyjVISmMZptD71BdClkvW6iWhk2yQlNCsiN8uJDslQfdlk6SdUCHR8LWKpUIsjr2I5Y29nt1zDEjUtuIYJemKzm9Wejn//H+M++9/QG3jZRvKYGQkv1ZCQdfEWtsoaYglB5IKE2NllaZ4df8BJGlaiK0fJ6+beygrsV+ba06hz+tpPknN+7q+qFLDCSrYfo0Us/KxkhgrmlD3WVWBv8E7zoDyyv798xdaRn7xi1/3uwlOj/gD3nEGlF8886t+N0Hx0d//3X43wekRf8A7zoDy+//0yn43QXH7bTv73QSnRwZGg2cPvQZpwInRAmN5Sa1ZFwdLimmsqam7TcGg0oqe4TZwICfbBj6u0nzTpFY5EatD8/eH/HcbGIrPsR3KzQGHTaCqsdE8WcTEyJq55fFRbfs8MZZvQyuvb/rotCqnNFHqFis5Tozlx10zTsGyKHkFAAxxkK1mXvfUtD7u5GTugcmmlg2rG3Mftcvzitr+Z71U9Ioql3SP1dGqKZjaiE2KUv49abbdd+y4C41GA+PjeZ9PT+fX2mrhyro4Ygppt6n7oULfFxFs2HAC9ux5Ca0sr2+4qb//BPUNo/w4IjrMnx7zcVF6KYJxNfibWUV1OkEM93OVV+rg4W/wjjOgnHjSCf1uguKtp7+l301wesQf8I4zoLztrZv73QTFT350f7+b4PRIHySazt+UeM5DTZVpWdXvenqdRMvpfcoDjNlAV9rwkE3O6plqFky6appD6f1oWptaMywyI4x4S9pzalLM8dHhsbnlcZZkAJxwfD49H23k5WzkQ/ZeZYnGBlpbd1xe//EnHD+3bCUkllvSRlyai5oAGpIa0/1QmI5H9ikEtCovVpRKygPGzXqh3nvvA0jTpooHz/JlQbKgdkySZGY9WWfa+TY20WU5tBDArpHiggvOxX33PaQlRevNKSy3qMbpMjwWk0g52Pu/ngii5CCb0zli4lx8zvTupRoinq1VwQSPsSOrv8E7zqDy2+cPlmPRnj37+t0Ep0fmfcCLyJtF5B4ReVxEHhORz3R/P15EdonI093/1x/75jrO6uHFvS/1uwmK557b0+8mOD1SR6JpAfjTEMKDIrIGwAMisgvAHwG4O4Rwk4hcD+B6AJ9baEP0F3f2Wox7slZNwet5tlppo9yaooqqYGMLoX7gsXLLIAu3j60VeNoPAENDuRfkmjVktUEp3QBghIKPHR49PLd89JCWaNoc2CwrlyIAYP1xa+eWjz8hf0cYGxtT5Viy4SqqLJeYVKwsxjuV93lmxlDd6zvb5yEEXZ8ZXryepuTR25Wj9u7dh0ajoTxUuR8KntN07uy5ayU8NQ4iljONIT0+Go0Grrrqw7jttp2QRiQwIKC9YVGOiETlzKWwotHBvPRxEkhkk657IWHEWB5cSJC0vqTsCyHsDSE82F1+HcATADYB2Abglm6xWwAMltGu47zB2XblZf1uguI2t4N/w9GTBi8imwGcB+A+ABtCCHu7m14EsCGyz3UisltEdrfaM2VFHMcp4fbv3tXvJihOPXVjv5vg9EjtB7yITAD4DoA/CSEc5G2hMx8pnZOEELaHELaEELY0TOowx3HibBqwB+rJJ5/Y7yY4PVLLTFJEmug83L8RQri1+/NLIrIxhLBXRDYCWNwndtLTdaS/uhXYgqzD9W4spLxBaxozVeluVdr6QrRF7qOG0ZczjrDI1mkqmqQ+5shQru2uW3McVabLsZfrkSNH5paPrtVafdbKBWaWaW0ij/Vr8/rWHb9ubtl6srI+zNEkhzPtSZkIfxNRG1S5Ot86rCnlfDlUy+rj62T1avX9hj1eu7r4xpM34KW9r2BkaHRu2+hw3s+tTM+IlWcrnW8xZy/vlK+MUDIX68maJAkefPBRJEmClK7hzIy+7tz/8fshi96Swd5r7DBc+3sXlSsch/u87Nc69c9P7Xt6IXlme6COFY0A+CqAJ0IIN9OmHQCu7S5fC+D2RbfGcZw5HvrZ4/1uguKCC87tdxOcHqnzavteAP8cwMUi8lD33+UAbgJwqYg8DeCS7rrjOEvElt8+p99NUDz77PP9boLTI/NKNCGEnyBu8fTBpW1OWQOqvD7LgzWVrfd8WJ4OLjBn57Eweyqru2CqFinH52QDVfE0nk0UC/LDRG7WOEUmlBwEC9AB0Nj61E79JybyY61Zlwcbsx/kWYYKSfw8YklgrExRR6KxZpIxYmMvhFDbjLbs2jz/3F4kSaL6bHg0l1EaJg+u6tuI6ScAtEJ+bYboOnEu3tSYSUojxSuvHYQ0UmVOXJQbVaJklNEpUyOhht2P96nIh6o9Wevdu4XULgt6fsSOVa+tReomVonjnqyOM6C88spr/W6C4sOXvLffTXB6pG/hgh1nNXNi4yR89LHLK8scvP4g1t60Fp9616ew5ZQt+OM7/xj3XHsPbvjBDXjqlaew+7rd2HTzJnz2PZ/FKWtOwZ99/8+w+1O7cd2d1+H1qddx5x/cibO+eBY+/4HPAwBu/OGNePLTT2Lr32zFmuE12L51O7Z8eQu+8KEvYM/re3DzvTfjhc++gC3bt+DMN52JGy68ARfdchG+tPVL2L1nN7b9h+oJ+4lDpZbSTh+RxUoZvTA2MhHO2NzRFW2wsUaaWwo0GjRVTHTQKZ5qj47l2zZu1IPr9DM2zy1v2nTy3PLQWF53c9hO7/Op59AIHbfKE5AsCmxc9gMHDswtczAua0nCFiMct9yWG0rLJ1x2mseyBbeVrzXnXQV0vPXWTL5PMYhYvszeqi0zC1Ueplk8hvnoaH7dx8Zy+eHo5BFVbmiI+oKqaBn30NAmaYjHdhLP8ZpFZJ0q2S/mITy77aPXbMV3vn2n+t3KBbzOnqO2j9grlePdZ5keb1xHzKMX0FZhfJ14Hzv2GhTgjeWfYh/FvME75a76vctx2x3fU6qFkhuNrMO5jNMq7+2sXBqSohlNadtt7P+2uqaxnL31np22rQt55l555YceCCFs6XU/l2gcx3FWKP6AdxzHWaH4A95xHGeFMjAfWZc6+lqVlko16H0q9LqFtKcqJ2WMWP7YKqwGXOdYYjN+RLZZLbfdYj2dkqI07fcMigiasTZsjkWRCVuBk0XYnL1sNkkmk0azjeX3tNEkeRtvCTXNT+uOV2X+aE8+Kz8WR5YErDbOWrj+PsX5gVm3t23l+mJafaVurKqzY6+0OvC9JiIm40y5d2lnXUqXrQlmor4LxNoQNxVeyPWsS9W3nOromW4m6TiO40TwB7zjOM4KZVklmiD5rNROTGKTkaw4Z8vrW2BwrzoETjoS9P4SceytG0TMBi+rnUgiYp4WjLlcyufOkk9V3WT/mFES1UKeyQZNcTlPaqqHkjqnNtvEmXOnaTJPs1muAQAhMzhWMKyUw/lfQyrRclxHRteXzfQSsw8HceO+tP06238BbZV31Zo/8jhg88KiBy3V0Si/tgDQJJls0piZMgHlbUqUCmNNP+m45RaE1fBOksXHYqUEUk+SVXlhzZOlrpmjDjQXSpbqt664D4+xeH3FnMC942/wjuM4K5SB+cjqOCuJwyeehGuu2dbvZgwcR046qd9NWFUs7wM+cCBxlXoaAAAJqUlEQVQmM/2gmdQMeSc2xMgPaW45wF5409PT+lA0/RoapfyeEp8KJ2QVEvPw6/xAUoLxXmU4f+aho/mUeeaIjqHNQaI4/2YjjUs+PB1PRQeGKvRtGaZMqzVdWmxo2FqfUFx2ntJbSYWtVNRpmNyX6nrkdTSbVrTj+uMT5eZQPcuDquuWHyUupSUVlkYJmvi7r3y9UF8q9W43K4upfq6a0kcsRGzgLJvHdhY+i4LcSPWpbYU0DLGxR/kLAKO9srd1XFRgCanQPlrmOgr3QnSbvdcWkENCnYdKxGDqzpcLllVEqBRV6+ESjeM4zgrFH/CO4zgrFH/AO47jrFCW/SPrrO5dCAaXUB7Wqv3JDq5Nepo1PWxHTN/SKjMpqo81OKt7KhO5ityt/C2hTXqfzaXJUQEzSsZgIzSqppKnqI12GUswsCSRQznZQ5XHIJk1JlVXNFJJoc95wFS6Ksb2Mbp2ZPdMi7nx41R8Bwih3JsWSU2zN2s7R31ZpcHzWbVpDGRWh6Y+V9FG2bzW9BDr30nFpQhZLJqkaQN3NOv2NuImexZHvWlNG7hcwc6a7knl9Wyi2yb8aCxPNFJ8fizk/qoYR5j/O9F8+Bu84zjOCsUf8I7jOCuUgbGD12ZP7FVpTBkj06BgpmLsK5e12Xysbhviv2tZRjVCl0tZ8qnIh0qrBc9Mrp7OnU4JDRM4LIlINFp9MF6adS2yOFAS1237KKs3XZXAEkH+e5rEh6ZUeSBGzOxMvg8TPCvfh4u1rTenkiYqgkTpvfI2GIkmJthY88ykwd7I1P9WIaADc8IO21adBKY8f6kY2U95DytzYi03cg8GdXfYZCfU7sq8qXS+NKaq7slMBXFDtBwnELHKEnuyZnSl9Ji3yUnKr6h9hjGhKuhfdEt9/A3ecRxnheIPeMdxnBXKsks0s1OkqikWf8FP7ERFTf3jf594GqrjeuenXB2LuSpYWVa6bGOscx7RkVGK1a0dWdFsUtAuuiI2Hnm7zdY2+bGGRedXrZoS5o0166F8pZAjkzzvssg+gI5pXulYG7kE1tuSvUXZkqpwbaQ8xndoxxshsX2MVyZbBgWOSW/jfUe8Oafbtpxam1uyclmbfuCAc2LGR6YkpCFajgdN421JxtYi+t5iFUXaJFG2jaUHD4qoV2uVRVcxv+1cdXydrAQYYtejIq49lzLj6OhRKz3ND1vD8Xi1QcOU/2ylx+zird78Dd5xHGeFMu8DXkRGROR+EXlYRB4TkRu7v58mIveJyDMi8j9FzGuk4ziO01fqvMFPAbg4hPBOAOcCuExE3g3gPwH4zyGE0wG8CuCTx66ZjuM4Tq/Mq8GHjmh1qLva7P4LAC4G8Afd328BcAOAv5qnNkhXP5VCpL5y7z9JjRbGnqgqN6fxDg1TtJxvk6Rc6+9s4+5gc0DbVjZVI+3USGZNNvsj4c1G9+OEGO2ZXMezETInJyfz1pFHKVqjuj6K0BjVOivM/KpMJvm7gMrtWXDqi+fZrEO7VWU+ViXql+e0TSo/BJS/59h+iOX5tS2N5dycmtHXM4to1Jk5v7hOG08Mokx57bVul2vw+ph6Hzbt5W9NrWl938UCTdoxz9t0nNCYJywKXq5qU0R3r/LertrWmir3Iq26N3iMKVPUqqwex5haGryIpCLyEIB9AHYB+AWA10KY86t/HsCmyL7XichuEdld+CDjOI7jHDNqPeBDCO0QwrkATgVwPoCz6x4ghLA9hLAlhLDFpnVzHMdxjh09PXFDCK+JyD0A3gPgOBFpdN/iTwXwwnz7T04defmxJ396GMDLC2ptL9xxzI+wFJyA5eiLwcf7oYP3Q473RYfZfnjLQnae9wEvIicCmOk+3EcBXIrOB9Z7AFwD4FsArgVw+3x1hRBOFJHdIYQtC2nsSsP7ooP3Qwfvhxzviw6L7Yc6b/AbAdwiHS+DBMD/CiHcKSKPA/iWiPx7AD8D8NWFNsJxHMdZeupY0TwC4LyS33+Jjh7vOI7jDCD98GTd3odjDireFx28Hzp4P+R4X3RYVD/IkmT5cRzHcQYOj0XjOI6zQlnWB7yIXCYiT3bj11y/nMfuJyLyZhG5R0Qe78bz+Uz39+NFZJeIPN39f32/27ocdB3nfiYid3bXV2VcIxE5TkS+LSI/F5EnROQ9q3FMiMi/6t4Xj4rIN7vxr1bFmBCRr4nIPhF5lH4rHQPS4S+6ffKIiLxrvvqX7QHftcL5SwAfAfB2AB8Xkbcv1/H7TAvAn4YQ3g7g3QD+RffcrwdwdwjhDAB3d9dXA58B8AStr9a4Rv8VwF0hhLMBvBOdPllVY0JENgH4lwC2hBDegU5KqI9h9YyJrwO4zPwWGwMfAXBG9991mDc0zPK+wZ8P4JkQwi9DCNPo2M9vW8bj940Qwt4QwoPd5dfRuZE3oXP+t3SL3QLgyv60cPkQkVMB/C6Ar3TXBZ24Rt/uFlkt/bAOwO+ga14cQpgOIbyGVTgm0LHmGxWRBoAxAHuxSsZECOFHAPabn2NjYBuAvw4d/i86zqYbq+pfzgf8JgDP0Xo0fs1KRkQ2o2N2eh+ADSGEvd1NLwLY0KdmLSf/BcC/Rh5j6k2oGddohXEagN8A+G9dueorIjKOVTYmQggvAPgCgP+HzoP9AIAHsDrHxCyxMdDzM9Q/si4jIjIB4DsA/iSEcJC3daN2rmiTJhHZCmBfCOGBfrdlAGgAeBeAvwohnAfgMIwcs0rGxHp03kxPA3AKgHEUJYtVy2LHwHI+4F8A8GZarxW/ZqUgIk10Hu7fCCHc2v35pdkpVvf/ff1q3zLxXgBXiMiz6Eh0F6OjQx/XnZ4Dq2dcPA/g+RDCfd31b6PzwF9tY+ISAL8KIfwmhDAD4FZ0xslqHBOzxMZAz8/Q5XzA/xTAGd2v40PofEjZsYzH7xtdnfmrAJ4IIdxMm3agE8cHqBnP541MCOHfhBBODSFsRuf6/30I4RPI4xoBq6AfACCE8CKA50TkrO5PHwTwOFbZmEBHmnm3iIx175PZflh1Y4KIjYEdAP6wa03zbgAHSMopJ4SwbP8AXA7gKXTiyf/b5Tx2P/8BeB8606xHADzU/Xc5Ovrz3QCeBvB3AI7vd1uXsU8uBHBnd/mtAO4H8AyA/w1guN/tW6Y+OBfA7u64+C6A9atxTAC4EcDPATwK4L8DGF4tYwLAN9H59jCDzqzuk7ExgE4mk7/sPj//AR3Lo8r63ZPVcRxnheIfWR3HcVYo/oB3HMdZofgD3nEcZ4XiD3jHcZwVij/gHcdxVij+gHccx1mh+APecRxnheIPeMdxnBXK/wfhG2RUct0KiQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "image = Image.open(path_to_image_file)\n",
    "plt.figure()\n",
    "currentAxis = plt.gca()\n",
    "currentAxis.imshow(image)\n",
    "currentAxis.add_patch(Rectangle((cropped_left, cropped_top), cropped_width, cropped_height, fill=False, edgecolor='red'))\n",
    "currentAxis.add_patch(Rectangle((bbox_left, bbox_top), bbox_width, bbox_height, fill=False, edgecolor='green'))\n",
    "for attr_left, attr_top, attr_width, attr_height in zip(attrs_left, attrs_top, attrs_width, attrs_height):\n",
    "    currentAxis.add_patch(Rectangle((attr_left, attr_top), attr_width, attr_height, fill=False, edgecolor='white', linestyle='dotted'))\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}