COCO_Image_Viewer.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# COCO Image Viewer\n",
    "This notebook will allow you to view details about a COCO dataset and preview segmentations on annotated images.\n",
    "Learn more about it at: http://cocodataset.org/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-07-27T08:58:05.441736Z",
     "start_time": "2019-07-27T08:58:05.075740Z"
    }
   },
   "outputs": [],
   "source": [
    "import IPython\n",
    "import os\n",
    "import json\n",
    "import random\n",
    "import numpy as np\n",
    "import requests\n",
    "from io import BytesIO\n",
    "from math import trunc\n",
    "from PIL import Image as PILImage\n",
    "from PIL import ImageDraw as PILImageDraw"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-07-27T08:58:05.486614Z",
     "start_time": "2019-07-27T08:58:05.443733Z"
    }
   },
   "outputs": [],
   "source": [
    "# Load the dataset json\n",
    "class CocoDataset():\n",
    "    def __init__(self, annotation_path, image_dir):\n",
    "        self.annotation_path = annotation_path\n",
    "        self.image_dir = image_dir\n",
    "        self.colors = ['blue', 'purple', 'red', 'green', 'orange', 'salmon', 'pink', 'gold',\n",
    "                       'orchid', 'slateblue', 'limegreen', 'seagreen', 'darkgreen', 'olive',\n",
    "                       'teal', 'aquamarine', 'steelblue', 'powderblue', 'dodgerblue', 'navy',\n",
    "                       'magenta', 'sienna', 'maroon']\n",
    "\n",
    "        json_file = open(self.annotation_path)\n",
    "        self.coco = json.load(json_file)\n",
    "        json_file.close()\n",
    "\n",
    "        self.process_info()\n",
    "        self.process_licenses()\n",
    "        self.process_categories()\n",
    "        self.process_images()\n",
    "        self.process_segmentations()\n",
    "\n",
    "    def display_info(self):\n",
    "        print('Dataset Info:')\n",
    "        print('=============')\n",
    "        if self.info is None:\n",
    "            return\n",
    "        for key, item in self.info.items():\n",
    "            print('  {}: {}'.format(key, item))\n",
    "\n",
    "        requirements = [['description', str],\n",
    "                        ['url', str],\n",
    "                        ['version', str],\n",
    "                        ['year', int],\n",
    "                        ['contributor', str],\n",
    "                        ['date_created', str]]\n",
    "        for req, req_type in requirements:\n",
    "            if req not in self.info:\n",
    "                print('ERROR: {} is missing'.format(req))\n",
    "            elif type(self.info[req]) != req_type:\n",
    "                print('ERROR: {} should be type {}'.format(req, str(req_type)))\n",
    "        print('')\n",
    "\n",
    "    def display_licenses(self):\n",
    "        print('Licenses:')\n",
    "        print('=========')\n",
    "\n",
    "        if self.licenses is None:\n",
    "            return\n",
    "        requirements = [['id', int],\n",
    "                        ['url', str],\n",
    "                        ['name', str]]\n",
    "        for license in self.licenses:\n",
    "            for key, item in license.items():\n",
    "                print('  {}: {}'.format(key, item))\n",
    "            for req, req_type in requirements:\n",
    "                if req not in license:\n",
    "                    print('ERROR: {} is missing'.format(req))\n",
    "                elif type(license[req]) != req_type:\n",
    "                    print('ERROR: {} should be type {}'.format(\n",
    "                        req, str(req_type)))\n",
    "            print('')\n",
    "        print('')\n",
    "\n",
    "    def display_categories(self):\n",
    "        print('Categories:')\n",
    "        print('=========')\n",
    "        for sc_key, sc_val in self.super_categories.items():\n",
    "            print('  super_category: {}'.format(sc_key))\n",
    "            for cat_id in sc_val:\n",
    "                print('    id {}: {}'.format(\n",
    "                    cat_id, self.categories[cat_id]['name']))\n",
    "            print('')\n",
    "\n",
    "    def display_image(self, image_id, show_polys=True, show_bbox=True, show_crowds=True, use_url=False):\n",
    "        print('Image:')\n",
    "        print('======')\n",
    "        if image_id == 'random':\n",
    "            image_id = random.choice(list(self.images.keys()))\n",
    "\n",
    "        # Print the image info\n",
    "        image = self.images[image_id]\n",
    "        for key, val in image.items():\n",
    "            print('  {}: {}'.format(key, val))\n",
    "\n",
    "        # Open the image\n",
    "        if use_url:\n",
    "            image_path = image['coco_url']\n",
    "            response = requests.get(image_path)\n",
    "            image = PILImage.open(BytesIO(response.content))\n",
    "\n",
    "        else:\n",
    "            # image_path = os.path.join(self.image_dir, image['file_name'])\n",
    "            image_path = \"{}/{}\".format(self.image_dir, image['file_name'])\n",
    "            image = PILImage.open(image_path)\n",
    "\n",
    "        # Calculate the size and adjusted display size\n",
    "        max_width = 600\n",
    "        image_width, image_height = image.size\n",
    "        adjusted_width = min(image_width, max_width)\n",
    "        adjusted_ratio = adjusted_width / image_width\n",
    "        adjusted_height = adjusted_ratio * image_height\n",
    "\n",
    "        # Create list of polygons to be drawn\n",
    "        polygons = {}\n",
    "        bbox_polygons = {}\n",
    "        rle_regions = {}\n",
    "        poly_colors = {}\n",
    "        bbox_categories = {}\n",
    "        print('  segmentations ({}):'.format(\n",
    "            len(self.segmentations[image_id])))\n",
    "        for i, segm in enumerate(self.segmentations[image_id]):\n",
    "            polygons_list = []\n",
    "            if segm['iscrowd'] != 0:\n",
    "                # Gotta decode the RLE\n",
    "                px = 0\n",
    "                x, y = 0, 0\n",
    "                rle_list = []\n",
    "                for j, counts in enumerate(segm['segmentation']['counts']):\n",
    "                    if j % 2 == 0:\n",
    "                        # Empty pixels\n",
    "                        px += counts\n",
    "                    else:\n",
    "                        # Need to draw on these pixels, since we are drawing in vector form,\n",
    "                        # we need to draw horizontal lines on the image\n",
    "                        x_start = trunc(\n",
    "                            trunc(px / image_height) * adjusted_ratio)\n",
    "                        y_start = trunc(px % image_height * adjusted_ratio)\n",
    "                        px += counts\n",
    "                        x_end = trunc(trunc(px / image_height)\n",
    "                                      * adjusted_ratio)\n",
    "                        y_end = trunc(px % image_height * adjusted_ratio)\n",
    "                        if x_end == x_start:\n",
    "                            # This is only on one line\n",
    "                            rle_list.append(\n",
    "                                {'x': x_start, 'y': y_start, 'width': 1, 'height': (y_end - y_start)})\n",
    "                        if x_end > x_start:\n",
    "                            # This spans more than one line\n",
    "                            # Insert top line first\n",
    "                            rle_list.append(\n",
    "                                {'x': x_start, 'y': y_start, 'width': 1, 'height': (image_height - y_start)})\n",
    "\n",
    "                            # Insert middle lines if needed\n",
    "                            lines_spanned = x_end - x_start + 1  # total number of lines spanned\n",
    "                            full_lines_to_insert = lines_spanned - 2\n",
    "                            if full_lines_to_insert > 0:\n",
    "                                full_lines_to_insert = trunc(\n",
    "                                    full_lines_to_insert * adjusted_ratio)\n",
    "                                rle_list.append(\n",
    "                                    {'x': (x_start + 1), 'y': 0, 'width': full_lines_to_insert, 'height': image_height})\n",
    "\n",
    "                            # Insert bottom line\n",
    "                            rle_list.append(\n",
    "                                {'x': x_end, 'y': 0, 'width': 1, 'height': y_end})\n",
    "                if len(rle_list) > 0:\n",
    "                    rle_regions[segm['id']] = rle_list\n",
    "            else:\n",
    "                # Add the polygon segmentation\n",
    "                for segmentation_points in segm['segmentation']:\n",
    "                    segmentation_points = np.multiply(\n",
    "                        segmentation_points, adjusted_ratio).astype(int)\n",
    "                    polygons_list.append(\n",
    "                        str(segmentation_points).lstrip('[').rstrip(']'))\n",
    "            polygons[segm['id']] = polygons_list\n",
    "            if i < len(self.colors):\n",
    "                poly_colors[segm['id']] = self.colors[i]\n",
    "            else:\n",
    "                poly_colors[segm['id']] = 'white'\n",
    "\n",
    "            bbox = segm['bbox']\n",
    "            bbox_points = [bbox[0], bbox[1], bbox[0] + bbox[2], bbox[1],\n",
    "                           bbox[0] + bbox[2], bbox[1] +\n",
    "                           bbox[3], bbox[0], bbox[1] + bbox[3],\n",
    "                           bbox[0], bbox[1]]\n",
    "            bbox_points = np.multiply(bbox_points, adjusted_ratio).astype(int)\n",
    "            bbox_polygons[segm['id']] = str(\n",
    "                bbox_points).lstrip('[').rstrip(']')\n",
    "            bbox_categories[segm['id']] = self.categories[segm['category_id']]\n",
    "            # Print details\n",
    "            print('    {}:{}:{}'.format(\n",
    "                segm['id'], poly_colors[segm['id']], self.categories[segm['category_id']]))\n",
    "\n",
    "        # Draw segmentation polygons on image\n",
    "        html = '<div class=\"container\" style=\"position:relative;\">'\n",
    "        html += '<img src=\"{}\" style=\"position:relative;top:0px;left:0px;width:{}px;\">'.format(\n",
    "            image_path, adjusted_width)\n",
    "        html += '<div class=\"svgclass\"><svg width=\"{}\" height=\"{}\">'.format(\n",
    "            adjusted_width, adjusted_height)\n",
    "\n",
    "        if show_polys:\n",
    "            for seg_id, points_list in polygons.items():\n",
    "                fill_color = poly_colors[seg_id]\n",
    "                stroke_color = poly_colors[seg_id]\n",
    "                for points in points_list:\n",
    "                    html += '<polygon points=\"{}\" style=\"fill:{}; stroke:{}; stroke-width:1; fill-opacity:0.5\" />'.format(\n",
    "                        points, fill_color, stroke_color)\n",
    "\n",
    "        if show_crowds:\n",
    "            for seg_id, rect_list in rle_regions.items():\n",
    "                fill_color = poly_colors[seg_id]\n",
    "                stroke_color = poly_colors[seg_id]\n",
    "                for rect_def in rect_list:\n",
    "                    x, y = rect_def['x'], rect_def['y']\n",
    "                    w, h = rect_def['width'], rect_def['height']\n",
    "                    html += '<rect x=\"{}\" y=\"{}\" width=\"{}\" height=\"{}\" style=\"fill:{}; stroke:{}; stroke-width:1; fill-opacity:0.5; stroke-opacity:0.5\" />'.format(\n",
    "                        x, y, w, h, fill_color, stroke_color)\n",
    "\n",
    "        if show_bbox:\n",
    "            for seg_id, points in bbox_polygons.items():\n",
    "                x, y = [int(i) for i in points.split()[:2]]\n",
    "                html += '<text x=\"{}\" y=\"{}\" fill=\"yellow\">{}</text>'.format(\n",
    "                    x, y, bbox_categories[seg_id][\"name\"])\n",
    "                fill_color = poly_colors[seg_id]\n",
    "                stroke_color = poly_colors[seg_id]\n",
    "                html += '<polygon points=\"{}\" style=\"fill:{}; stroke:{}; stroke-width:1; fill-opacity:0\" />'.format(\n",
    "                    points, fill_color, stroke_color)\n",
    "\n",
    "        html += '</svg></div>'\n",
    "        html += '</div>'\n",
    "        html += '<style>'\n",
    "        html += '.svgclass { position:absolute; top:0px; left:0px;}'\n",
    "        html += '</style>'\n",
    "        return html\n",
    "\n",
    "    def process_info(self):\n",
    "        self.info = self.coco.get('info')\n",
    "\n",
    "    def process_licenses(self):\n",
    "        self.licenses = self.coco.get('licenses')\n",
    "\n",
    "    def process_categories(self):\n",
    "        self.categories = {}\n",
    "        self.super_categories = {}\n",
    "        for category in self.coco['categories']:\n",
    "            cat_id = category['id']\n",
    "            super_category = category['supercategory']\n",
    "\n",
    "            # Add category to the categories dict\n",
    "            if cat_id not in self.categories:\n",
    "                self.categories[cat_id] = category\n",
    "            else:\n",
    "                print(\"ERROR: Skipping duplicate category id: {}\".format(category))\n",
    "\n",
    "            # Add category to super_categories dict\n",
    "            if super_category not in self.super_categories:\n",
    "                # Create a new set with the category id\n",
    "                self.super_categories[super_category] = {cat_id}\n",
    "            else:\n",
    "                self.super_categories[super_category] |= {\n",
    "                    cat_id}  # Add category id to the set\n",
    "\n",
    "    def process_images(self):\n",
    "        self.images = {}\n",
    "        for image in self.coco['images']:\n",
    "            image_id = image['id']\n",
    "            if image_id in self.images:\n",
    "                print(\"ERROR: Skipping duplicate image id: {}\".format(image))\n",
    "            else:\n",
    "                self.images[image_id] = image\n",
    "\n",
    "    def process_segmentations(self):\n",
    "        self.segmentations = {}\n",
    "        for segmentation in self.coco['annotations']:\n",
    "            image_id = segmentation['image_id']\n",
    "            if image_id not in self.segmentations:\n",
    "                self.segmentations[image_id] = []\n",
    "            self.segmentations[image_id].append(segmentation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-07-27T08:58:05.510585Z",
     "start_time": "2019-07-27T08:58:05.489608Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset Info:\n",
      "=============\n",
      "Licenses:\n",
      "=========\n",
      "Categories:\n",
      "=========\n",
      "  super_category: date\n",
      "    id 1: date\n",
      "\n",
      "  super_category: fig\n",
      "    id 2: fig\n",
      "\n",
      "  super_category: hazelnut\n",
      "    id 3: hazelnut\n",
      "\n"
     ]
    }
   ],
   "source": [
    "annotation_path = \"trainval.json\"\n",
    "image_dir = \"images\"\n",
    "\n",
    "coco_dataset = CocoDataset(annotation_path, image_dir)\n",
    "coco_dataset.display_info()\n",
    "coco_dataset.display_licenses()\n",
    "coco_dataset.display_categories()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-07-27T08:58:05.520552Z",
     "start_time": "2019-07-27T08:58:05.512547Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 {'height': 600, 'width': 800, 'id': 0, 'file_name': '0.jpg'}\n",
      "1 {'height': 600, 'width': 800, 'id': 1, 'file_name': '1.jpg'}\n",
      "2 {'height': 600, 'width': 800, 'id': 2, 'file_name': '10.jpg'}\n",
      "3 {'height': 600, 'width': 800, 'id': 3, 'file_name': '11.jpg'}\n",
      "4 {'height': 600, 'width': 800, 'id': 4, 'file_name': '12.jpg'}\n",
      "5 {'height': 600, 'width': 800, 'id': 5, 'file_name': '13.jpg'}\n",
      "6 {'height': 600, 'width': 800, 'id': 6, 'file_name': '14.jpg'}\n",
      "7 {'height': 600, 'width': 800, 'id': 7, 'file_name': '15.jpg'}\n",
      "8 {'height': 600, 'width': 800, 'id': 8, 'file_name': '16.jpg'}\n",
      "9 {'height': 600, 'width': 800, 'id': 9, 'file_name': '17.jpg'}\n",
      "10 {'height': 600, 'width': 800, 'id': 10, 'file_name': '2.jpg'}\n",
      "11 {'height': 600, 'width': 800, 'id': 11, 'file_name': '3.jpg'}\n",
      "12 {'height': 600, 'width': 800, 'id': 12, 'file_name': '4.jpg'}\n",
      "13 {'height': 600, 'width': 800, 'id': 13, 'file_name': '5.jpg'}\n",
      "14 {'height': 600, 'width': 800, 'id': 14, 'file_name': '6.jpg'}\n",
      "15 {'height': 600, 'width': 800, 'id': 15, 'file_name': '7.jpg'}\n",
      "16 {'height': 600, 'width': 800, 'id': 16, 'file_name': '8.jpg'}\n",
      "17 {'height': 600, 'width': 800, 'id': 17, 'file_name': '9.jpg'}\n"
     ]
    }
   ],
   "source": [
    "for k, v in coco_dataset.images.items():\n",
    "    print(k, v)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-07-27T08:58:05.572387Z",
     "start_time": "2019-07-27T08:58:05.523519Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Image:\n",
      "======\n",
      "  height: 600\n",
      "  width: 800\n",
      "  id: 5\n",
      "  file_name: 13.jpg\n",
      "  segmentations (8):\n",
      "    53:blue:{'supercategory': 'date', 'id': 1, 'name': 'date'}\n",
      "    54:purple:{'supercategory': 'date', 'id': 1, 'name': 'date'}\n",
      "    55:red:{'supercategory': 'date', 'id': 1, 'name': 'date'}\n",
      "    56:green:{'supercategory': 'hazelnut', 'id': 3, 'name': 'hazelnut'}\n",
      "    57:orange:{'supercategory': 'hazelnut', 'id': 3, 'name': 'hazelnut'}\n",
      "    58:salmon:{'supercategory': 'hazelnut', 'id': 3, 'name': 'hazelnut'}\n",
      "    59:pink:{'supercategory': 'hazelnut', 'id': 3, 'name': 'hazelnut'}\n",
      "    60:gold:{'supercategory': 'fig', 'id': 2, 'name': 'fig'}\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div class=\"container\" style=\"position:relative;\"><img src=\"images/13.jpg\" style=\"position:relative;top:0px;left:0px;width:600px;\"><div class=\"svgclass\"><svg width=\"600\" height=\"450.0\"><polygon points=\"190  43 190  50 187  84 202 107 227 119 256 123 289 128 316 121 317 105\n",
       " 324  84 311  71 288  56 252  29 237  21 200  25 189  36\" style=\"fill:blue; stroke:blue; stroke-width:1; fill-opacity:0.5\" /><polygon points=\" 38 111  87  94 109 101 148 120 176 131 178 142 174 168 164 178 149 181\n",
       " 108 181  85 183  55 178  33 151  32 130\" style=\"fill:purple; stroke:purple; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"331 145 299 173 298 195 324 219 352 223 383 210 434 173 438 153 430 134\n",
       " 407 132 380 139\" style=\"fill:red; stroke:red; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"138 193 165 193 185 199 193 215 199 217 195 220 181 242 156 248 133 241\n",
       " 118 217 126 200\" style=\"fill:green; stroke:green; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"359 242 373 236 397 238 406 242 410 254 411 268 397 278 367 277 354 267\n",
       " 352 261 354 251\" style=\"fill:orange; stroke:orange; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"456 273 439 285 437 298 449 315 476 329 499 326 505 318 503 313 502 293\n",
       " 488 283 469 272\" style=\"fill:salmon; stroke:salmon; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"404 111 409 120 435 129 455 124 461 112 465 108 461 107 458  91 424  77\n",
       " 411  80 404  91 403  96\" style=\"fill:pink; stroke:pink; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"430 334 479 328 492 336 486 351 471 368 457 389 438 399 413 391 401 374\n",
       " 408 351\" style=\"fill:gold; stroke:gold; stroke-width:1; fill-opacity:0.5\" /><text x=\"186\" y=\"21\" fill=\"yellow\">date</text><polygon points=\"186  21 324  21 324 128 186 128 186  21\" style=\"fill:blue; stroke:blue; stroke-width:1; fill-opacity:0\" /><text x=\"31\" y=\"94\" fill=\"yellow\">date</text><polygon points=\" 31  94 177  94 177 183  31 183  31  94\" style=\"fill:purple; stroke:purple; stroke-width:1; fill-opacity:0\" /><text x=\"297\" y=\"132\" fill=\"yellow\">date</text><polygon points=\"297 132 438 132 438 222 297 222 297 132\" style=\"fill:red; stroke:red; stroke-width:1; fill-opacity:0\" /><text x=\"118\" y=\"193\" fill=\"yellow\">hazelnut</text><polygon points=\"118 193 199 193 199 247 118 247 118 193\" style=\"fill:green; stroke:green; stroke-width:1; fill-opacity:0\" /><text x=\"351\" y=\"236\" fill=\"yellow\">hazelnut</text><polygon points=\"351 236 411 236 411 278 351 278 351 236\" style=\"fill:orange; stroke:orange; stroke-width:1; fill-opacity:0\" /><text x=\"437\" y=\"271\" fill=\"yellow\">hazelnut</text><polygon points=\"437 271 505 271 505 328 437 328 437 271\" style=\"fill:salmon; stroke:salmon; stroke-width:1; fill-opacity:0\" /><text x=\"402\" y=\"76\" fill=\"yellow\">hazelnut</text><polygon points=\"402  76 465  76 465 129 402 129 402  76\" style=\"fill:pink; stroke:pink; stroke-width:1; fill-opacity:0\" /><text x=\"401\" y=\"327\" fill=\"yellow\">fig</text><polygon points=\"401 327 492 327 492 399 401 399 401 327\" style=\"fill:gold; stroke:gold; stroke-width:1; fill-opacity:0\" /></svg></div></div><style>.svgclass { position:absolute; top:0px; left:0px;}</style>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "html = coco_dataset.display_image(5, use_url=False)\n",
    "IPython.display.HTML(html)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-07-27T08:58:05.587373Z",
     "start_time": "2019-07-27T08:58:05.574383Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Image:\n",
      "======\n",
      "  height: 600\n",
      "  width: 800\n",
      "  id: 1\n",
      "  file_name: 1.jpg\n",
      "  segmentations (9):\n",
      "    13:blue:{'supercategory': 'hazelnut', 'id': 3, 'name': 'hazelnut'}\n",
      "    14:purple:{'supercategory': 'hazelnut', 'id': 3, 'name': 'hazelnut'}\n",
      "    15:red:{'supercategory': 'hazelnut', 'id': 3, 'name': 'hazelnut'}\n",
      "    16:green:{'supercategory': 'date', 'id': 1, 'name': 'date'}\n",
      "    17:orange:{'supercategory': 'date', 'id': 1, 'name': 'date'}\n",
      "    18:salmon:{'supercategory': 'date', 'id': 1, 'name': 'date'}\n",
      "    19:pink:{'supercategory': 'date', 'id': 1, 'name': 'date'}\n",
      "    20:gold:{'supercategory': 'date', 'id': 1, 'name': 'date'}\n",
      "    21:orchid:{'supercategory': 'fig', 'id': 2, 'name': 'fig'}\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div class=\"container\" style=\"position:relative;\"><img src=\"images/1.jpg\" style=\"position:relative;top:0px;left:0px;width:600px;\"><div class=\"svgclass\"><svg width=\"600\" height=\"450.0\"><polygon points=\"230  74 258  56 261  35 264  26 255  23 248  15 233  10 208  24 199  43\n",
       " 200  58 211  70\" style=\"fill:blue; stroke:blue; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"465  93 469 121 486 141 504 146 516 151 520 138 532 118 530 100 515  75\n",
       " 498  71 477  80\" style=\"fill:purple; stroke:purple; stroke-width:1; fill-opacity:0.5\" /><polygon points=\" 59  92  68  98  80 106  86 135  75 157  51 163  34 150  27 118  45  98\" style=\"fill:red; stroke:red; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"125 257  90 283  72 358  80 377  98 385 118 385 178 335 190 299 178 268\n",
       " 166 268 162 261\" style=\"fill:green; stroke:green; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"284 205 206 266 198 315 217 335 256 346 285 334 307 291 312 283 336 229\n",
       " 329 218 305 200\" style=\"fill:orange; stroke:orange; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"375 155 443 145 463 150 475 148 483 163 486 161 488 181 484 194 454 210\n",
       " 410 241 379 242 346 234 336 204 348 166\" style=\"fill:salmon; stroke:salmon; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"522 237 455 305 452 342 470 369 495 378 528 371 552 338 553 323 580 271\n",
       " 577 245 551 231\" style=\"fill:pink; stroke:pink; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"309 321 265 379 266 407 293 419 323 414 363 410 393 378 392 351 385 344\n",
       " 384 329 362 309 346 305 321 314\" style=\"fill:gold; stroke:gold; stroke-width:1; fill-opacity:0.5\" /><polygon points=\"293  94 292 125 303 136 333 147 352 136 378  91 379  58 384  45 374  44\n",
       " 362  53 330  58 306  69\" style=\"fill:orchid; stroke:orchid; stroke-width:1; fill-opacity:0.5\" /><text x=\"199\" y=\"9\" fill=\"yellow\">hazelnut</text><polygon points=\"199   9 264   9 264  74 199  74 199   9\" style=\"fill:blue; stroke:blue; stroke-width:1; fill-opacity:0\" /><text x=\"465\" y=\"71\" fill=\"yellow\">hazelnut</text><polygon points=\"465  71 531  71 531 150 465 150 465  71\" style=\"fill:purple; stroke:purple; stroke-width:1; fill-opacity:0\" /><text x=\"27\" y=\"92\" fill=\"yellow\">hazelnut</text><polygon points=\" 27  92  85  92  85 163  27 163  27  92\" style=\"fill:red; stroke:red; stroke-width:1; fill-opacity:0\" /><text x=\"72\" y=\"257\" fill=\"yellow\">date</text><polygon points=\" 72 257 189 257 189 384  72 384  72 257\" style=\"fill:green; stroke:green; stroke-width:1; fill-opacity:0\" /><text x=\"198\" y=\"200\" fill=\"yellow\">date</text><polygon points=\"198 200 336 200 336 346 198 346 198 200\" style=\"fill:orange; stroke:orange; stroke-width:1; fill-opacity:0\" /><text x=\"336\" y=\"145\" fill=\"yellow\">date</text><polygon points=\"336 145 487 145 487 242 336 242 336 145\" style=\"fill:salmon; stroke:salmon; stroke-width:1; fill-opacity:0\" /><text x=\"452\" y=\"231\" fill=\"yellow\">date</text><polygon points=\"452 231 580 231 580 378 452 378 452 231\" style=\"fill:pink; stroke:pink; stroke-width:1; fill-opacity:0\" /><text x=\"264\" y=\"305\" fill=\"yellow\">date</text><polygon points=\"264 305 393 305 393 419 264 419 264 305\" style=\"fill:gold; stroke:gold; stroke-width:1; fill-opacity:0\" /><text x=\"291\" y=\"44\" fill=\"yellow\">fig</text><polygon points=\"291  44 384  44 384 147 291 147 291  44\" style=\"fill:orchid; stroke:orchid; stroke-width:1; fill-opacity:0\" /></svg></div></div><style>.svgclass { position:absolute; top:0px; left:0px;}</style>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "html = coco_dataset.display_image(1, use_url=False)\n",
    "IPython.display.HTML(html)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}