├── LICENSE
├── README.md
├── basic_data_loading.ipynb
├── data
    ├── annotations_classify_task.pkl
    ├── annotations_contains_task.pkl
    ├── image_info.pkl
    ├── imagenet_label_map.npy
    ├── imagenet_label_to_wnid.npy
    ├── model_predictions.pkl
    └── superclasses.npy
└── pipeline.jpg


/LICENSE:
--------------------------------------------------------------------------------
 1 | MIT License
 2 | 
 3 | Copyright (c) 2020 Madry Lab
 4 | 
 5 | Permission is hereby granted, free of charge, to any person obtaining a copy
 6 | of this software and associated documentation files (the "Software"), to deal
 7 | in the Software without restriction, including without limitation the rights
 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 9 | copies of the Software, and to permit persons to whom the Software is
10 | furnished to do so, subject to the following conditions:
11 | 
12 | The above copyright notice and this permission notice shall be included in all
13 | copies or substantial portions of the Software.
14 | 
15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 | SOFTWARE.
22 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
 1 | # Fine-grained annotations for the ImageNet validation set
 2 | This is the data collected for our paper "From ImageNet to Image Classification: Contextualizing Progress on Benchmarks" ([preprint](https://arxiv.org/abs/2005.11295), [blog](https://gradientscience.org/benchmarks)).
 3 | 
 4 | ![](pipeline.jpg)
 5 | 
 6 | ## Parsing the annotations
 7 | Our annotations are available as `pandas` dataframes in `data/annotations_{contains,classify}_task.pkl`. These dataframes included both the raw data collected (after quality control), as well as the aggregate quantities we computed for our analysis. The rest of the files in `data` contain auxiliary information.
 8 | 
 9 | The easiest way to navigate these files is by running the jupyter notebook (`basic_data_loading.ipynb`) which loads all files, providing an explanation for each field.
10 | 
11 | ## Citation
12 | 
13 | ```
14 | @inproceedings{tsipras2020imagenet,
15 |     title={From ImageNet to Image Classification: Contextualizing Progress on Benchmarks},
16 |     author={Dimitris Tsipras and Shibani Santurkar and Logan Engstrom and Andrew Ilyas and Aleksander Madry},
17 |     booktitle={ArXiv preprint arXiv:2005.11295},
18 |     year={2020}
19 | }
20 | ```
21 | 
22 | 


--------------------------------------------------------------------------------
/basic_data_loading.ipynb:
--------------------------------------------------------------------------------
   1 | {
   2 |  "cells": [
   3 |   {
   4 |    "cell_type": "markdown",
   5 |    "metadata": {},
   6 |    "source": [
   7 |     "# Load and visualize data"
   8 |    ]
   9 |   },
  10 |   {
  11 |    "cell_type": "code",
  12 |    "execution_count": 1,
  13 |    "metadata": {},
  14 |    "outputs": [],
  15 |    "source": [
  16 |     "import pandas as pd\n",
  17 |     "import numpy as np\n",
  18 |     "from PIL import Image\n",
  19 |     "import matplotlib.pyplot as plt\n",
  20 |     "\n",
  21 |     "%matplotlib inline"
  22 |    ]
  23 |   },
  24 |   {
  25 |    "cell_type": "markdown",
  26 |    "metadata": {},
  27 |    "source": [
  28 |     "#### ImageNet label map"
  29 |    ]
  30 |   },
  31 |   {
  32 |    "cell_type": "code",
  33 |    "execution_count": 2,
  34 |    "metadata": {},
  35 |    "outputs": [],
  36 |    "source": [
  37 |     "label_map = np.load('./data/imagenet_label_map.npy', allow_pickle=True).item()\n",
  38 |     "label_to_wnid = np.load('./data/imagenet_label_to_wnid.npy', allow_pickle=True).item()"
  39 |    ]
  40 |   },
  41 |   {
  42 |    "cell_type": "markdown",
  43 |    "metadata": {},
  44 |    "source": [
  45 |     "### ImageNet superclasses"
  46 |    ]
  47 |   },
  48 |   {
  49 |    "cell_type": "code",
  50 |    "execution_count": 3,
  51 |    "metadata": {},
  52 |    "outputs": [
  53 |     {
  54 |      "name": "stdout",
  55 |      "output_type": "stream",
  56 |      "text": [
  57 |       "Dogs, #ImageNet classes: 130, Sample classes: [151, 152, 153, 154, 155]\n",
  58 |       "Other mammals, #ImageNet classes: 88, Sample classes: [286, 287, 288, 289, 290]\n",
  59 |       "Birds, #ImageNet classes: 59, Sample classes: [128, 129, 130, 131, 132]\n",
  60 |       "Reptiles, fish, amphibians, #ImageNet classes: 60, Sample classes: [33, 34, 35, 36, 37]\n",
  61 |       "inverterbrates, #ImageNet classes: 61, Sample classes: [300, 301, 302, 303, 304]\n",
  62 |       "Food, plants, fungi, #ImageNet classes: 63, Sample classes: [992, 993, 994, 995, 996]\n",
  63 |       "Devices, #ImageNet classes: 172, Sample classes: [513, 517, 527, 530, 531]\n",
  64 |       "Structures, furnishing, #ImageNet classes: 90, Sample classes: [516, 648, 520, 526, 532]\n",
  65 |       "Clothes, covering, #ImageNet classes: 92, Sample classes: [643, 515, 518, 775, 903]\n",
  66 |       "Implements, containers, misc. objects, #ImageNet classes: 117, Sample classes: [512, 644, 521, 523, 909]\n",
  67 |       "vehicles, #ImageNet classes: 68, Sample classes: [779, 780, 654, 913, 914]\n"
  68 |      ]
  69 |     }
  70 |    ],
  71 |    "source": [
  72 |     "superclasses = np.load('./data/superclasses.npy', allow_pickle=True).item()\n",
  73 |     "\n",
  74 |     "for k, v in superclasses.items():\n",
  75 |     "    print(f\"{k}, #ImageNet classes: {len(v)}, Sample classes: {v[:5]}\")"
  76 |    ]
  77 |   },
  78 |   {
  79 |    "cell_type": "markdown",
  80 |    "metadata": {},
  81 |    "source": [
  82 |     "## Model predictions\n",
  83 |     "\n",
  84 |     "The dataframe contains the following information:\n",
  85 |     "\n",
  86 |     "____________________________________________________________________________________________________\n",
  87 |     "\n",
  88 |     "`image_number`: number of image as per an unshuffled loader\n",
  89 |     "\n",
  90 |     "`imagenet_label`: ImageNet label\n",
  91 |     "\n",
  92 |     "____________________________________________________________________________________________________\n",
  93 |     "\n",
  94 |     "`pred_{MODEL_NAME}`: Top-5 predictions of model\n",
  95 |     "\n",
  96 |     "`top1_{MODEL_NAME}`: True if label == Top-1 model prediction\n",
  97 |     "\n",
  98 |     "`top5_{MODEL_NAME}`: True if label is in Top-5 model predictions"
  99 |    ]
 100 |   },
 101 |   {
 102 |    "cell_type": "code",
 103 |    "execution_count": 4,
 104 |    "metadata": {},
 105 |    "outputs": [],
 106 |    "source": [
 107 |     "model_preds = pd.read_pickle(\"./data/model_predictions.pkl\")"
 108 |    ]
 109 |   },
 110 |   {
 111 |    "cell_type": "code",
 112 |    "execution_count": 5,
 113 |    "metadata": {},
 114 |    "outputs": [
 115 |     {
 116 |      "data": {
 117 |       "text/html": [
 118 |        "<div>\n",
 119 |        "<style scoped>\n",
 120 |        "    .dataframe tbody tr th:only-of-type {\n",
 121 |        "        vertical-align: middle;\n",
 122 |        "    }\n",
 123 |        "\n",
 124 |        "    .dataframe tbody tr th {\n",
 125 |        "        vertical-align: top;\n",
 126 |        "    }\n",
 127 |        "\n",
 128 |        "    .dataframe thead th {\n",
 129 |        "        text-align: right;\n",
 130 |        "    }\n",
 131 |        "</style>\n",
 132 |        "<table border=\"1\" class=\"dataframe\">\n",
 133 |        "  <thead>\n",
 134 |        "    <tr style=\"text-align: right;\">\n",
 135 |        "      <th></th>\n",
 136 |        "      <th>img_number</th>\n",
 137 |        "      <th>imagenet_label</th>\n",
 138 |        "      <th>pred_googlenet</th>\n",
 139 |        "      <th>pred_alexnet</th>\n",
 140 |        "      <th>pred_resnet50</th>\n",
 141 |        "      <th>pred_resnet101</th>\n",
 142 |        "      <th>pred_inception_v3</th>\n",
 143 |        "      <th>pred_vgg16</th>\n",
 144 |        "      <th>pred_wide_resnet50_2</th>\n",
 145 |        "      <th>pred_densenet161</th>\n",
 146 |        "      <th>...</th>\n",
 147 |        "      <th>top5_googlenet</th>\n",
 148 |        "      <th>top5_alexnet</th>\n",
 149 |        "      <th>top5_resnet50</th>\n",
 150 |        "      <th>top5_resnet101</th>\n",
 151 |        "      <th>top5_inception_v3</th>\n",
 152 |        "      <th>top5_vgg16</th>\n",
 153 |        "      <th>top5_wide_resnet50_2</th>\n",
 154 |        "      <th>top5_densenet161</th>\n",
 155 |        "      <th>top5_mobilenet_v2</th>\n",
 156 |        "      <th>top5_tf_efficientnet_b7</th>\n",
 157 |        "    </tr>\n",
 158 |        "  </thead>\n",
 159 |        "  <tbody>\n",
 160 |        "    <tr>\n",
 161 |        "      <th>0</th>\n",
 162 |        "      <td>0</td>\n",
 163 |        "      <td>0</td>\n",
 164 |        "      <td>[48, 39, 0, 391, 394]</td>\n",
 165 |        "      <td>[395, 0, 48, 390, 394]</td>\n",
 166 |        "      <td>[394, 0, 395, 391, 389]</td>\n",
 167 |        "      <td>[394, 0, 149, 395, 391]</td>\n",
 168 |        "      <td>[0, 394, 48, 36, 391]</td>\n",
 169 |        "      <td>[395, 394, 391, 48, 39]</td>\n",
 170 |        "      <td>[394, 391, 0, 389, 390]</td>\n",
 171 |        "      <td>[394, 0, 391, 48, 389]</td>\n",
 172 |        "      <td>...</td>\n",
 173 |        "      <td>True</td>\n",
 174 |        "      <td>True</td>\n",
 175 |        "      <td>True</td>\n",
 176 |        "      <td>True</td>\n",
 177 |        "      <td>True</td>\n",
 178 |        "      <td>False</td>\n",
 179 |        "      <td>True</td>\n",
 180 |        "      <td>True</td>\n",
 181 |        "      <td>True</td>\n",
 182 |        "      <td>True</td>\n",
 183 |        "    </tr>\n",
 184 |        "    <tr>\n",
 185 |        "      <th>1</th>\n",
 186 |        "      <td>1</td>\n",
 187 |        "      <td>0</td>\n",
 188 |        "      <td>[0, 391, 122, 766, 389]</td>\n",
 189 |        "      <td>[764, 763, 413, 758, 391]</td>\n",
 190 |        "      <td>[0, 389, 391, 395, 758]</td>\n",
 191 |        "      <td>[0, 389, 391, 758, 394]</td>\n",
 192 |        "      <td>[391, 0, 758, 389, 764]</td>\n",
 193 |        "      <td>[0, 391, 758, 389, 395]</td>\n",
 194 |        "      <td>[0, 758, 391, 389, 395]</td>\n",
 195 |        "      <td>[0, 391, 389, 758, 395]</td>\n",
 196 |        "      <td>...</td>\n",
 197 |        "      <td>True</td>\n",
 198 |        "      <td>False</td>\n",
 199 |        "      <td>True</td>\n",
 200 |        "      <td>True</td>\n",
 201 |        "      <td>True</td>\n",
 202 |        "      <td>True</td>\n",
 203 |        "      <td>True</td>\n",
 204 |        "      <td>True</td>\n",
 205 |        "      <td>True</td>\n",
 206 |        "      <td>True</td>\n",
 207 |        "    </tr>\n",
 208 |        "    <tr>\n",
 209 |        "      <th>2</th>\n",
 210 |        "      <td>2</td>\n",
 211 |        "      <td>0</td>\n",
 212 |        "      <td>[0, 391, 346, 693, 223]</td>\n",
 213 |        "      <td>[929, 472, 444, 870, 655]</td>\n",
 214 |        "      <td>[0, 391, 693, 389, 390]</td>\n",
 215 |        "      <td>[0, 693, 758, 391, 472]</td>\n",
 216 |        "      <td>[837, 703, 419, 836, 785]</td>\n",
 217 |        "      <td>[0, 758, 391, 223, 237]</td>\n",
 218 |        "      <td>[0, 693, 391, 472, 389]</td>\n",
 219 |        "      <td>[0, 391, 395, 693, 472]</td>\n",
 220 |        "      <td>...</td>\n",
 221 |        "      <td>True</td>\n",
 222 |        "      <td>False</td>\n",
 223 |        "      <td>True</td>\n",
 224 |        "      <td>True</td>\n",
 225 |        "      <td>False</td>\n",
 226 |        "      <td>True</td>\n",
 227 |        "      <td>True</td>\n",
 228 |        "      <td>True</td>\n",
 229 |        "      <td>True</td>\n",
 230 |        "      <td>True</td>\n",
 231 |        "    </tr>\n",
 232 |        "    <tr>\n",
 233 |        "      <th>3</th>\n",
 234 |        "      <td>3</td>\n",
 235 |        "      <td>0</td>\n",
 236 |        "      <td>[0, 391, 389, 394, 395]</td>\n",
 237 |        "      <td>[0, 389, 395, 36, 955]</td>\n",
 238 |        "      <td>[0, 389, 391, 395, 394]</td>\n",
 239 |        "      <td>[0, 389, 391, 758, 997]</td>\n",
 240 |        "      <td>[0, 391, 122, 955, 933]</td>\n",
 241 |        "      <td>[0, 389, 30, 758, 390]</td>\n",
 242 |        "      <td>[0, 389, 395, 955, 36]</td>\n",
 243 |        "      <td>[0, 389, 391, 394, 397]</td>\n",
 244 |        "      <td>...</td>\n",
 245 |        "      <td>True</td>\n",
 246 |        "      <td>True</td>\n",
 247 |        "      <td>True</td>\n",
 248 |        "      <td>True</td>\n",
 249 |        "      <td>True</td>\n",
 250 |        "      <td>True</td>\n",
 251 |        "      <td>True</td>\n",
 252 |        "      <td>True</td>\n",
 253 |        "      <td>True</td>\n",
 254 |        "      <td>True</td>\n",
 255 |        "    </tr>\n",
 256 |        "    <tr>\n",
 257 |        "      <th>4</th>\n",
 258 |        "      <td>4</td>\n",
 259 |        "      <td>0</td>\n",
 260 |        "      <td>[0, 391, 1, 389, 556]</td>\n",
 261 |        "      <td>[0, 389, 114, 390, 939]</td>\n",
 262 |        "      <td>[0, 389, 391, 758, 955]</td>\n",
 263 |        "      <td>[0, 955, 389, 382, 1]</td>\n",
 264 |        "      <td>[0, 389, 37, 265, 5]</td>\n",
 265 |        "      <td>[0, 389, 758, 391, 390]</td>\n",
 266 |        "      <td>[0, 389, 758, 391, 395]</td>\n",
 267 |        "      <td>[0, 389, 904, 955, 391]</td>\n",
 268 |        "      <td>...</td>\n",
 269 |        "      <td>True</td>\n",
 270 |        "      <td>True</td>\n",
 271 |        "      <td>True</td>\n",
 272 |        "      <td>True</td>\n",
 273 |        "      <td>True</td>\n",
 274 |        "      <td>True</td>\n",
 275 |        "      <td>True</td>\n",
 276 |        "      <td>True</td>\n",
 277 |        "      <td>True</td>\n",
 278 |        "      <td>True</td>\n",
 279 |        "    </tr>\n",
 280 |        "    <tr>\n",
 281 |        "      <th>...</th>\n",
 282 |        "      <td>...</td>\n",
 283 |        "      <td>...</td>\n",
 284 |        "      <td>...</td>\n",
 285 |        "      <td>...</td>\n",
 286 |        "      <td>...</td>\n",
 287 |        "      <td>...</td>\n",
 288 |        "      <td>...</td>\n",
 289 |        "      <td>...</td>\n",
 290 |        "      <td>...</td>\n",
 291 |        "      <td>...</td>\n",
 292 |        "      <td>...</td>\n",
 293 |        "      <td>...</td>\n",
 294 |        "      <td>...</td>\n",
 295 |        "      <td>...</td>\n",
 296 |        "      <td>...</td>\n",
 297 |        "      <td>...</td>\n",
 298 |        "      <td>...</td>\n",
 299 |        "      <td>...</td>\n",
 300 |        "      <td>...</td>\n",
 301 |        "      <td>...</td>\n",
 302 |        "      <td>...</td>\n",
 303 |        "    </tr>\n",
 304 |        "    <tr>\n",
 305 |        "      <th>49995</th>\n",
 306 |        "      <td>49995</td>\n",
 307 |        "      <td>999</td>\n",
 308 |        "      <td>[896, 799, 435, 876, 706]</td>\n",
 309 |        "      <td>[799, 905, 876, 435, 831]</td>\n",
 310 |        "      <td>[896, 905, 904, 435, 799]</td>\n",
 311 |        "      <td>[896, 876, 435, 804, 861]</td>\n",
 312 |        "      <td>[876, 435, 579, 532, 526]</td>\n",
 313 |        "      <td>[799, 896, 435, 876, 904]</td>\n",
 314 |        "      <td>[999, 700, 904, 896, 861]</td>\n",
 315 |        "      <td>[896, 700, 435, 876, 904]</td>\n",
 316 |        "      <td>...</td>\n",
 317 |        "      <td>False</td>\n",
 318 |        "      <td>False</td>\n",
 319 |        "      <td>False</td>\n",
 320 |        "      <td>False</td>\n",
 321 |        "      <td>False</td>\n",
 322 |        "      <td>False</td>\n",
 323 |        "      <td>True</td>\n",
 324 |        "      <td>False</td>\n",
 325 |        "      <td>False</td>\n",
 326 |        "      <td>True</td>\n",
 327 |        "    </tr>\n",
 328 |        "    <tr>\n",
 329 |        "      <th>49996</th>\n",
 330 |        "      <td>49996</td>\n",
 331 |        "      <td>999</td>\n",
 332 |        "      <td>[999, 700, 868, 861, 435]</td>\n",
 333 |        "      <td>[725, 504, 876, 666, 896]</td>\n",
 334 |        "      <td>[999, 700, 876, 435, 868]</td>\n",
 335 |        "      <td>[700, 999, 599, 876, 934]</td>\n",
 336 |        "      <td>[700, 999, 896, 435, 731]</td>\n",
 337 |        "      <td>[999, 700, 861, 896, 868]</td>\n",
 338 |        "      <td>[999, 876, 896, 435, 804]</td>\n",
 339 |        "      <td>[999, 700, 868, 876, 965]</td>\n",
 340 |        "      <td>...</td>\n",
 341 |        "      <td>True</td>\n",
 342 |        "      <td>False</td>\n",
 343 |        "      <td>True</td>\n",
 344 |        "      <td>True</td>\n",
 345 |        "      <td>True</td>\n",
 346 |        "      <td>True</td>\n",
 347 |        "      <td>True</td>\n",
 348 |        "      <td>True</td>\n",
 349 |        "      <td>True</td>\n",
 350 |        "      <td>True</td>\n",
 351 |        "    </tr>\n",
 352 |        "    <tr>\n",
 353 |        "      <th>49997</th>\n",
 354 |        "      <td>49997</td>\n",
 355 |        "      <td>999</td>\n",
 356 |        "      <td>[762, 532, 495, 896, 857]</td>\n",
 357 |        "      <td>[470, 857, 532, 854, 406]</td>\n",
 358 |        "      <td>[854, 406, 851, 492, 762]</td>\n",
 359 |        "      <td>[521, 505, 968, 899, 859]</td>\n",
 360 |        "      <td>[857, 854, 406, 498, 762]</td>\n",
 361 |        "      <td>[406, 745, 762, 470, 896]</td>\n",
 362 |        "      <td>[495, 532, 725, 505, 968]</td>\n",
 363 |        "      <td>[406, 982, 607, 470, 846]</td>\n",
 364 |        "      <td>...</td>\n",
 365 |        "      <td>False</td>\n",
 366 |        "      <td>False</td>\n",
 367 |        "      <td>False</td>\n",
 368 |        "      <td>False</td>\n",
 369 |        "      <td>False</td>\n",
 370 |        "      <td>False</td>\n",
 371 |        "      <td>False</td>\n",
 372 |        "      <td>False</td>\n",
 373 |        "      <td>False</td>\n",
 374 |        "      <td>True</td>\n",
 375 |        "    </tr>\n",
 376 |        "    <tr>\n",
 377 |        "      <th>49998</th>\n",
 378 |        "      <td>49998</td>\n",
 379 |        "      <td>999</td>\n",
 380 |        "      <td>[333, 937, 700, 330, 999]</td>\n",
 381 |        "      <td>[333, 937, 356, 359, 334]</td>\n",
 382 |        "      <td>[333, 700, 937, 338, 356]</td>\n",
 383 |        "      <td>[333, 700, 999, 937, 794]</td>\n",
 384 |        "      <td>[333, 937, 700, 338, 999]</td>\n",
 385 |        "      <td>[333, 937, 987, 338, 357]</td>\n",
 386 |        "      <td>[333, 937, 961, 700, 338]</td>\n",
 387 |        "      <td>[333, 700, 999, 728, 937]</td>\n",
 388 |        "      <td>...</td>\n",
 389 |        "      <td>True</td>\n",
 390 |        "      <td>False</td>\n",
 391 |        "      <td>False</td>\n",
 392 |        "      <td>True</td>\n",
 393 |        "      <td>True</td>\n",
 394 |        "      <td>False</td>\n",
 395 |        "      <td>False</td>\n",
 396 |        "      <td>True</td>\n",
 397 |        "      <td>False</td>\n",
 398 |        "      <td>False</td>\n",
 399 |        "    </tr>\n",
 400 |        "    <tr>\n",
 401 |        "      <th>49999</th>\n",
 402 |        "      <td>49999</td>\n",
 403 |        "      <td>999</td>\n",
 404 |        "      <td>[999, 700, 772, 728, 845]</td>\n",
 405 |        "      <td>[474, 850, 911, 658, 700]</td>\n",
 406 |        "      <td>[999, 700, 543, 605, 772]</td>\n",
 407 |        "      <td>[700, 999, 605, 728, 455]</td>\n",
 408 |        "      <td>[605, 700, 999, 761, 633]</td>\n",
 409 |        "      <td>[714, 543, 999, 605, 805]</td>\n",
 410 |        "      <td>[999, 700, 419, 605, 574]</td>\n",
 411 |        "      <td>[999, 700, 728, 543, 723]</td>\n",
 412 |        "      <td>...</td>\n",
 413 |        "      <td>True</td>\n",
 414 |        "      <td>False</td>\n",
 415 |        "      <td>True</td>\n",
 416 |        "      <td>True</td>\n",
 417 |        "      <td>True</td>\n",
 418 |        "      <td>True</td>\n",
 419 |        "      <td>True</td>\n",
 420 |        "      <td>True</td>\n",
 421 |        "      <td>True</td>\n",
 422 |        "      <td>True</td>\n",
 423 |        "    </tr>\n",
 424 |        "  </tbody>\n",
 425 |        "</table>\n",
 426 |        "<p>50000 rows × 32 columns</p>\n",
 427 |        "</div>"
 428 |       ],
 429 |       "text/plain": [
 430 |        "       img_number  imagenet_label             pred_googlenet  \\\n",
 431 |        "0               0               0      [48, 39, 0, 391, 394]   \n",
 432 |        "1               1               0    [0, 391, 122, 766, 389]   \n",
 433 |        "2               2               0    [0, 391, 346, 693, 223]   \n",
 434 |        "3               3               0    [0, 391, 389, 394, 395]   \n",
 435 |        "4               4               0      [0, 391, 1, 389, 556]   \n",
 436 |        "...           ...             ...                        ...   \n",
 437 |        "49995       49995             999  [896, 799, 435, 876, 706]   \n",
 438 |        "49996       49996             999  [999, 700, 868, 861, 435]   \n",
 439 |        "49997       49997             999  [762, 532, 495, 896, 857]   \n",
 440 |        "49998       49998             999  [333, 937, 700, 330, 999]   \n",
 441 |        "49999       49999             999  [999, 700, 772, 728, 845]   \n",
 442 |        "\n",
 443 |        "                    pred_alexnet              pred_resnet50  \\\n",
 444 |        "0         [395, 0, 48, 390, 394]    [394, 0, 395, 391, 389]   \n",
 445 |        "1      [764, 763, 413, 758, 391]    [0, 389, 391, 395, 758]   \n",
 446 |        "2      [929, 472, 444, 870, 655]    [0, 391, 693, 389, 390]   \n",
 447 |        "3         [0, 389, 395, 36, 955]    [0, 389, 391, 395, 394]   \n",
 448 |        "4        [0, 389, 114, 390, 939]    [0, 389, 391, 758, 955]   \n",
 449 |        "...                          ...                        ...   \n",
 450 |        "49995  [799, 905, 876, 435, 831]  [896, 905, 904, 435, 799]   \n",
 451 |        "49996  [725, 504, 876, 666, 896]  [999, 700, 876, 435, 868]   \n",
 452 |        "49997  [470, 857, 532, 854, 406]  [854, 406, 851, 492, 762]   \n",
 453 |        "49998  [333, 937, 356, 359, 334]  [333, 700, 937, 338, 356]   \n",
 454 |        "49999  [474, 850, 911, 658, 700]  [999, 700, 543, 605, 772]   \n",
 455 |        "\n",
 456 |        "                  pred_resnet101          pred_inception_v3  \\\n",
 457 |        "0        [394, 0, 149, 395, 391]      [0, 394, 48, 36, 391]   \n",
 458 |        "1        [0, 389, 391, 758, 394]    [391, 0, 758, 389, 764]   \n",
 459 |        "2        [0, 693, 758, 391, 472]  [837, 703, 419, 836, 785]   \n",
 460 |        "3        [0, 389, 391, 758, 997]    [0, 391, 122, 955, 933]   \n",
 461 |        "4          [0, 955, 389, 382, 1]       [0, 389, 37, 265, 5]   \n",
 462 |        "...                          ...                        ...   \n",
 463 |        "49995  [896, 876, 435, 804, 861]  [876, 435, 579, 532, 526]   \n",
 464 |        "49996  [700, 999, 599, 876, 934]  [700, 999, 896, 435, 731]   \n",
 465 |        "49997  [521, 505, 968, 899, 859]  [857, 854, 406, 498, 762]   \n",
 466 |        "49998  [333, 700, 999, 937, 794]  [333, 937, 700, 338, 999]   \n",
 467 |        "49999  [700, 999, 605, 728, 455]  [605, 700, 999, 761, 633]   \n",
 468 |        "\n",
 469 |        "                      pred_vgg16       pred_wide_resnet50_2  \\\n",
 470 |        "0        [395, 394, 391, 48, 39]    [394, 391, 0, 389, 390]   \n",
 471 |        "1        [0, 391, 758, 389, 395]    [0, 758, 391, 389, 395]   \n",
 472 |        "2        [0, 758, 391, 223, 237]    [0, 693, 391, 472, 389]   \n",
 473 |        "3         [0, 389, 30, 758, 390]     [0, 389, 395, 955, 36]   \n",
 474 |        "4        [0, 389, 758, 391, 390]    [0, 389, 758, 391, 395]   \n",
 475 |        "...                          ...                        ...   \n",
 476 |        "49995  [799, 896, 435, 876, 904]  [999, 700, 904, 896, 861]   \n",
 477 |        "49996  [999, 700, 861, 896, 868]  [999, 876, 896, 435, 804]   \n",
 478 |        "49997  [406, 745, 762, 470, 896]  [495, 532, 725, 505, 968]   \n",
 479 |        "49998  [333, 937, 987, 338, 357]  [333, 937, 961, 700, 338]   \n",
 480 |        "49999  [714, 543, 999, 605, 805]  [999, 700, 419, 605, 574]   \n",
 481 |        "\n",
 482 |        "                pred_densenet161  ... top5_googlenet top5_alexnet  \\\n",
 483 |        "0         [394, 0, 391, 48, 389]  ...           True         True   \n",
 484 |        "1        [0, 391, 389, 758, 395]  ...           True        False   \n",
 485 |        "2        [0, 391, 395, 693, 472]  ...           True        False   \n",
 486 |        "3        [0, 389, 391, 394, 397]  ...           True         True   \n",
 487 |        "4        [0, 389, 904, 955, 391]  ...           True         True   \n",
 488 |        "...                          ...  ...            ...          ...   \n",
 489 |        "49995  [896, 700, 435, 876, 904]  ...          False        False   \n",
 490 |        "49996  [999, 700, 868, 876, 965]  ...           True        False   \n",
 491 |        "49997  [406, 982, 607, 470, 846]  ...          False        False   \n",
 492 |        "49998  [333, 700, 999, 728, 937]  ...           True        False   \n",
 493 |        "49999  [999, 700, 728, 543, 723]  ...           True        False   \n",
 494 |        "\n",
 495 |        "       top5_resnet50  top5_resnet101  top5_inception_v3  top5_vgg16  \\\n",
 496 |        "0               True            True               True       False   \n",
 497 |        "1               True            True               True        True   \n",
 498 |        "2               True            True              False        True   \n",
 499 |        "3               True            True               True        True   \n",
 500 |        "4               True            True               True        True   \n",
 501 |        "...              ...             ...                ...         ...   \n",
 502 |        "49995          False           False              False       False   \n",
 503 |        "49996           True            True               True        True   \n",
 504 |        "49997          False           False              False       False   \n",
 505 |        "49998          False            True               True       False   \n",
 506 |        "49999           True            True               True        True   \n",
 507 |        "\n",
 508 |        "       top5_wide_resnet50_2  top5_densenet161  top5_mobilenet_v2  \\\n",
 509 |        "0                      True              True               True   \n",
 510 |        "1                      True              True               True   \n",
 511 |        "2                      True              True               True   \n",
 512 |        "3                      True              True               True   \n",
 513 |        "4                      True              True               True   \n",
 514 |        "...                     ...               ...                ...   \n",
 515 |        "49995                  True             False              False   \n",
 516 |        "49996                  True              True               True   \n",
 517 |        "49997                 False             False              False   \n",
 518 |        "49998                 False              True              False   \n",
 519 |        "49999                  True              True               True   \n",
 520 |        "\n",
 521 |        "       top5_tf_efficientnet_b7  \n",
 522 |        "0                         True  \n",
 523 |        "1                         True  \n",
 524 |        "2                         True  \n",
 525 |        "3                         True  \n",
 526 |        "4                         True  \n",
 527 |        "...                        ...  \n",
 528 |        "49995                     True  \n",
 529 |        "49996                     True  \n",
 530 |        "49997                     True  \n",
 531 |        "49998                    False  \n",
 532 |        "49999                     True  \n",
 533 |        "\n",
 534 |        "[50000 rows x 32 columns]"
 535 |       ]
 536 |      },
 537 |      "execution_count": 5,
 538 |      "metadata": {},
 539 |      "output_type": "execute_result"
 540 |     }
 541 |    ],
 542 |    "source": [
 543 |     "model_preds"
 544 |    ]
 545 |   },
 546 |   {
 547 |    "cell_type": "markdown",
 548 |    "metadata": {},
 549 |    "source": [
 550 |     "## CONTAINS task with multiple labels/image\n",
 551 |     "\n",
 552 |     "The dataframe contains the following information:\n",
 553 |     "\n",
 554 |     "____________________________________________________________________________________________________\n",
 555 |     "\n",
 556 |     "`image`: image filename \n",
 557 |     "\n",
 558 |     "`image_number`: number of image as per an unshuffled loader (same as above)\n",
 559 |     "\n",
 560 |     "`imagenet_label`: ImageNet label\n",
 561 |     "\n",
 562 |     "____________________________________________________________________________________________________\n",
 563 |     "\n",
 564 |     "`query`: set of candidate labels queried in \"contains\" experiment\n",
 565 |     "\n",
 566 |     "`num_queries`: size of `query` set\n",
 567 |     "\n",
 568 |     "____________________________________________________________________________________________________\n",
 569 |     "\n",
 570 |     "`selected`: number of annotators that selected each query label\n",
 571 |     "\n",
 572 |     "`total`: number of annotators that saw each query label (some HITs were discarded as part of quality control)\n",
 573 |     "\n",
 574 |     "`sel_dict`: dictionary mapping query label to tuple with number of workers that selected the image-query pair, and the number of workers that it was presented to "
 575 |    ]
 576 |   },
 577 |   {
 578 |    "cell_type": "code",
 579 |    "execution_count": 6,
 580 |    "metadata": {},
 581 |    "outputs": [],
 582 |    "source": [
 583 |     "info_contains = pd.read_pickle(\"./data/annotations_contains_task.pkl\")"
 584 |    ]
 585 |   },
 586 |   {
 587 |    "cell_type": "code",
 588 |    "execution_count": 7,
 589 |    "metadata": {},
 590 |    "outputs": [
 591 |     {
 592 |      "data": {
 593 |       "text/html": [
 594 |        "<div>\n",
 595 |        "<style scoped>\n",
 596 |        "    .dataframe tbody tr th:only-of-type {\n",
 597 |        "        vertical-align: middle;\n",
 598 |        "    }\n",
 599 |        "\n",
 600 |        "    .dataframe tbody tr th {\n",
 601 |        "        vertical-align: top;\n",
 602 |        "    }\n",
 603 |        "\n",
 604 |        "    .dataframe thead th {\n",
 605 |        "        text-align: right;\n",
 606 |        "    }\n",
 607 |        "</style>\n",
 608 |        "<table border=\"1\" class=\"dataframe\">\n",
 609 |        "  <thead>\n",
 610 |        "    <tr style=\"text-align: right;\">\n",
 611 |        "      <th></th>\n",
 612 |        "      <th>Unnamed: 0</th>\n",
 613 |        "      <th>query</th>\n",
 614 |        "      <th>selected</th>\n",
 615 |        "      <th>total</th>\n",
 616 |        "      <th>image_number</th>\n",
 617 |        "      <th>imagenet_label</th>\n",
 618 |        "      <th>num_queries</th>\n",
 619 |        "      <th>sel_dict</th>\n",
 620 |        "    </tr>\n",
 621 |        "    <tr>\n",
 622 |        "      <th>image</th>\n",
 623 |        "      <th></th>\n",
 624 |        "      <th></th>\n",
 625 |        "      <th></th>\n",
 626 |        "      <th></th>\n",
 627 |        "      <th></th>\n",
 628 |        "      <th></th>\n",
 629 |        "      <th></th>\n",
 630 |        "      <th></th>\n",
 631 |        "    </tr>\n",
 632 |        "  </thead>\n",
 633 |        "  <tbody>\n",
 634 |        "    <tr>\n",
 635 |        "      <th>ILSVRC2012_val_00047683.JPEG</th>\n",
 636 |        "      <td>90113</td>\n",
 637 |        "      <td>[392, 310, 322, 712, 321, 325, 946, 309, 326, ...</td>\n",
 638 |        "      <td>[0, 0, 4, 0, 4, 7, 1, 0, 6, 1, 4, 0]</td>\n",
 639 |        "      <td>[9, 9, 9, 7, 5, 8, 9, 9, 7, 9, 9, 9]</td>\n",
 640 |        "      <td>16296</td>\n",
 641 |        "      <td>325</td>\n",
 642 |        "      <td>12</td>\n",
 643 |        "      <td>{392: (0, 9), 310: (0, 9), 322: (4, 9), 712: (...</td>\n",
 644 |        "    </tr>\n",
 645 |        "    <tr>\n",
 646 |        "      <th>ILSVRC2012_val_00030990.JPEG</th>\n",
 647 |        "      <td>46639</td>\n",
 648 |        "      <td>[312, 30, 39, 31, 149, 46, 40, 47, 42, 55, 401]</td>\n",
 649 |        "      <td>[0, 0, 8, 0, 1, 3, 4, 7, 2, 1, 0]</td>\n",
 650 |        "      <td>[7, 9, 8, 8, 8, 7, 8, 8, 4, 9, 8]</td>\n",
 651 |        "      <td>2382</td>\n",
 652 |        "      <td>47</td>\n",
 653 |        "      <td>11</td>\n",
 654 |        "      <td>{312: (0, 7), 30: (0, 9), 39: (8, 8), 31: (0, ...</td>\n",
 655 |        "    </tr>\n",
 656 |        "    <tr>\n",
 657 |        "      <th>ILSVRC2012_val_00002431.JPEG</th>\n",
 658 |        "      <td>67588</td>\n",
 659 |        "      <td>[870, 444, 670, 826, 778, 685, 63, 880, 597, 4...</td>\n",
 660 |        "      <td>[1, 0, 0, 0, 1, 0, 0, 2, 0, 0, 1, 7, 1, 0, 0, ...</td>\n",
 661 |        "      <td>[9, 9, 8, 9, 8, 9, 7, 8, 9, 8, 9, 9, 7, 9, 9, ...</td>\n",
 662 |        "      <td>35201</td>\n",
 663 |        "      <td>704</td>\n",
 664 |        "      <td>17</td>\n",
 665 |        "      <td>{870: (1, 9), 444: (0, 9), 670: (0, 8), 826: (...</td>\n",
 666 |        "    </tr>\n",
 667 |        "    <tr>\n",
 668 |        "      <th>ILSVRC2012_val_00033061.JPEG</th>\n",
 669 |        "      <td>125421</td>\n",
 670 |        "      <td>[497, 442, 448, 494, 884, 698, 406, 425, 663, ...</td>\n",
 671 |        "      <td>[6, 8, 0, 4, 0, 0, 1, 1, 4, 1, 1]</td>\n",
 672 |        "      <td>[6, 8, 9, 9, 8, 7, 8, 9, 6, 8, 9]</td>\n",
 673 |        "      <td>22133</td>\n",
 674 |        "      <td>442</td>\n",
 675 |        "      <td>11</td>\n",
 676 |        "      <td>{497: (6, 6), 442: (8, 8), 448: (0, 9), 494: (...</td>\n",
 677 |        "    </tr>\n",
 678 |        "    <tr>\n",
 679 |        "      <th>ILSVRC2012_val_00002735.JPEG</th>\n",
 680 |        "      <td>126066</td>\n",
 681 |        "      <td>[497, 442, 884, 698, 399, 668, 832, 663, 538]</td>\n",
 682 |        "      <td>[6, 3, 4, 4, 0, 6, 6, 5, 8]</td>\n",
 683 |        "      <td>[8, 8, 8, 9, 8, 7, 8, 8, 9]</td>\n",
 684 |        "      <td>33403</td>\n",
 685 |        "      <td>668</td>\n",
 686 |        "      <td>9</td>\n",
 687 |        "      <td>{497: (6, 8), 442: (3, 8), 884: (4, 8), 698: (...</td>\n",
 688 |        "    </tr>\n",
 689 |        "    <tr>\n",
 690 |        "      <th>...</th>\n",
 691 |        "      <td>...</td>\n",
 692 |        "      <td>...</td>\n",
 693 |        "      <td>...</td>\n",
 694 |        "      <td>...</td>\n",
 695 |        "      <td>...</td>\n",
 696 |        "      <td>...</td>\n",
 697 |        "      <td>...</td>\n",
 698 |        "      <td>...</td>\n",
 699 |        "    </tr>\n",
 700 |        "    <tr>\n",
 701 |        "      <th>ILSVRC2012_val_00003714.JPEG</th>\n",
 702 |        "      <td>143137</td>\n",
 703 |        "      <td>[214, 177, 170, 231, 236, 264, 224, 205, 240, ...</td>\n",
 704 |        "      <td>[3, 1, 2, 1, 2, 1, 1, 2, 4, 1, 1, 1, 1, 1, 0, ...</td>\n",
 705 |        "      <td>[9, 6, 8, 9, 8, 6, 5, 8, 8, 8, 8, 7, 9, 8, 8, ...</td>\n",
 706 |        "      <td>10702</td>\n",
 707 |        "      <td>214</td>\n",
 708 |        "      <td>19</td>\n",
 709 |        "      <td>{214: (3, 9), 177: (1, 6), 170: (2, 8), 231: (...</td>\n",
 710 |        "    </tr>\n",
 711 |        "    <tr>\n",
 712 |        "      <th>ILSVRC2012_val_00018015.JPEG</th>\n",
 713 |        "      <td>54180</td>\n",
 714 |        "      <td>[843, 671, 892, 871, 814, 826, 778, 915, 880, ...</td>\n",
 715 |        "      <td>[0, 2, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 0, 0, ...</td>\n",
 716 |        "      <td>[9, 9, 9, 8, 9, 9, 5, 8, 9, 8, 7, 5, 9, 6, 9, ...</td>\n",
 717 |        "      <td>33564</td>\n",
 718 |        "      <td>671</td>\n",
 719 |        "      <td>23</td>\n",
 720 |        "      <td>{843: (0, 9), 671: (2, 9), 892: (1, 9), 871: (...</td>\n",
 721 |        "    </tr>\n",
 722 |        "    <tr>\n",
 723 |        "      <th>ILSVRC2012_val_00029538.JPEG</th>\n",
 724 |        "      <td>129790</td>\n",
 725 |        "      <td>[610, 714, 566, 745, 771, 515, 592, 650, 420, ...</td>\n",
 726 |        "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 8, 0, 0]</td>\n",
 727 |        "      <td>[9, 9, 9, 8, 6, 9, 9, 8, 9, 8, 8, 8, 9]</td>\n",
 728 |        "      <td>20129</td>\n",
 729 |        "      <td>402</td>\n",
 730 |        "      <td>13</td>\n",
 731 |        "      <td>{610: (0, 9), 714: (0, 9), 566: (0, 9), 745: (...</td>\n",
 732 |        "    </tr>\n",
 733 |        "    <tr>\n",
 734 |        "      <th>ILSVRC2012_val_00040262.JPEG</th>\n",
 735 |        "      <td>96869</td>\n",
 736 |        "      <td>[794, 440, 868, 585, 709, 611, 118, 917, 857, ...</td>\n",
 737 |        "      <td>[0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 1, 0, 6, 0, 0, ...</td>\n",
 738 |        "      <td>[9, 9, 8, 7, 8, 9, 8, 9, 9, 7, 8, 8, 8, 8, 8, ...</td>\n",
 739 |        "      <td>46093</td>\n",
 740 |        "      <td>921</td>\n",
 741 |        "      <td>19</td>\n",
 742 |        "      <td>{794: (0, 9), 440: (0, 9), 868: (0, 8), 585: (...</td>\n",
 743 |        "    </tr>\n",
 744 |        "    <tr>\n",
 745 |        "      <th>ILSVRC2012_val_00006124.JPEG</th>\n",
 746 |        "      <td>20850</td>\n",
 747 |        "      <td>[758, 726, 589, 618, 606, 666, 910, 828, 813, ...</td>\n",
 748 |        "      <td>[1, 1, 0, 2, 0, 0, 2, 5, 0, 0, 0, 2, 4, 0, 0, ...</td>\n",
 749 |        "      <td>[8, 8, 8, 9, 9, 8, 9, 9, 9, 7, 8, 8, 9, 9, 9, ...</td>\n",
 750 |        "      <td>36454</td>\n",
 751 |        "      <td>729</td>\n",
 752 |        "      <td>18</td>\n",
 753 |        "      <td>{758: (1, 8), 726: (1, 8), 589: (0, 8), 618: (...</td>\n",
 754 |        "    </tr>\n",
 755 |        "  </tbody>\n",
 756 |        "</table>\n",
 757 |        "<p>10000 rows × 8 columns</p>\n",
 758 |        "</div>"
 759 |       ],
 760 |       "text/plain": [
 761 |        "                              Unnamed: 0  \\\n",
 762 |        "image                                      \n",
 763 |        "ILSVRC2012_val_00047683.JPEG       90113   \n",
 764 |        "ILSVRC2012_val_00030990.JPEG       46639   \n",
 765 |        "ILSVRC2012_val_00002431.JPEG       67588   \n",
 766 |        "ILSVRC2012_val_00033061.JPEG      125421   \n",
 767 |        "ILSVRC2012_val_00002735.JPEG      126066   \n",
 768 |        "...                                  ...   \n",
 769 |        "ILSVRC2012_val_00003714.JPEG      143137   \n",
 770 |        "ILSVRC2012_val_00018015.JPEG       54180   \n",
 771 |        "ILSVRC2012_val_00029538.JPEG      129790   \n",
 772 |        "ILSVRC2012_val_00040262.JPEG       96869   \n",
 773 |        "ILSVRC2012_val_00006124.JPEG       20850   \n",
 774 |        "\n",
 775 |        "                                                                          query  \\\n",
 776 |        "image                                                                             \n",
 777 |        "ILSVRC2012_val_00047683.JPEG  [392, 310, 322, 712, 321, 325, 946, 309, 326, ...   \n",
 778 |        "ILSVRC2012_val_00030990.JPEG    [312, 30, 39, 31, 149, 46, 40, 47, 42, 55, 401]   \n",
 779 |        "ILSVRC2012_val_00002431.JPEG  [870, 444, 670, 826, 778, 685, 63, 880, 597, 4...   \n",
 780 |        "ILSVRC2012_val_00033061.JPEG  [497, 442, 448, 494, 884, 698, 406, 425, 663, ...   \n",
 781 |        "ILSVRC2012_val_00002735.JPEG      [497, 442, 884, 698, 399, 668, 832, 663, 538]   \n",
 782 |        "...                                                                         ...   \n",
 783 |        "ILSVRC2012_val_00003714.JPEG  [214, 177, 170, 231, 236, 264, 224, 205, 240, ...   \n",
 784 |        "ILSVRC2012_val_00018015.JPEG  [843, 671, 892, 871, 814, 826, 778, 915, 880, ...   \n",
 785 |        "ILSVRC2012_val_00029538.JPEG  [610, 714, 566, 745, 771, 515, 592, 650, 420, ...   \n",
 786 |        "ILSVRC2012_val_00040262.JPEG  [794, 440, 868, 585, 709, 611, 118, 917, 857, ...   \n",
 787 |        "ILSVRC2012_val_00006124.JPEG  [758, 726, 589, 618, 606, 666, 910, 828, 813, ...   \n",
 788 |        "\n",
 789 |        "                                                                       selected  \\\n",
 790 |        "image                                                                             \n",
 791 |        "ILSVRC2012_val_00047683.JPEG               [0, 0, 4, 0, 4, 7, 1, 0, 6, 1, 4, 0]   \n",
 792 |        "ILSVRC2012_val_00030990.JPEG                  [0, 0, 8, 0, 1, 3, 4, 7, 2, 1, 0]   \n",
 793 |        "ILSVRC2012_val_00002431.JPEG  [1, 0, 0, 0, 1, 0, 0, 2, 0, 0, 1, 7, 1, 0, 0, ...   \n",
 794 |        "ILSVRC2012_val_00033061.JPEG                  [6, 8, 0, 4, 0, 0, 1, 1, 4, 1, 1]   \n",
 795 |        "ILSVRC2012_val_00002735.JPEG                        [6, 3, 4, 4, 0, 6, 6, 5, 8]   \n",
 796 |        "...                                                                         ...   \n",
 797 |        "ILSVRC2012_val_00003714.JPEG  [3, 1, 2, 1, 2, 1, 1, 2, 4, 1, 1, 1, 1, 1, 0, ...   \n",
 798 |        "ILSVRC2012_val_00018015.JPEG  [0, 2, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 0, 0, ...   \n",
 799 |        "ILSVRC2012_val_00029538.JPEG            [0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 8, 0, 0]   \n",
 800 |        "ILSVRC2012_val_00040262.JPEG  [0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 1, 0, 6, 0, 0, ...   \n",
 801 |        "ILSVRC2012_val_00006124.JPEG  [1, 1, 0, 2, 0, 0, 2, 5, 0, 0, 0, 2, 4, 0, 0, ...   \n",
 802 |        "\n",
 803 |        "                                                                          total  \\\n",
 804 |        "image                                                                             \n",
 805 |        "ILSVRC2012_val_00047683.JPEG               [9, 9, 9, 7, 5, 8, 9, 9, 7, 9, 9, 9]   \n",
 806 |        "ILSVRC2012_val_00030990.JPEG                  [7, 9, 8, 8, 8, 7, 8, 8, 4, 9, 8]   \n",
 807 |        "ILSVRC2012_val_00002431.JPEG  [9, 9, 8, 9, 8, 9, 7, 8, 9, 8, 9, 9, 7, 9, 9, ...   \n",
 808 |        "ILSVRC2012_val_00033061.JPEG                  [6, 8, 9, 9, 8, 7, 8, 9, 6, 8, 9]   \n",
 809 |        "ILSVRC2012_val_00002735.JPEG                        [8, 8, 8, 9, 8, 7, 8, 8, 9]   \n",
 810 |        "...                                                                         ...   \n",
 811 |        "ILSVRC2012_val_00003714.JPEG  [9, 6, 8, 9, 8, 6, 5, 8, 8, 8, 8, 7, 9, 8, 8, ...   \n",
 812 |        "ILSVRC2012_val_00018015.JPEG  [9, 9, 9, 8, 9, 9, 5, 8, 9, 8, 7, 5, 9, 6, 9, ...   \n",
 813 |        "ILSVRC2012_val_00029538.JPEG            [9, 9, 9, 8, 6, 9, 9, 8, 9, 8, 8, 8, 9]   \n",
 814 |        "ILSVRC2012_val_00040262.JPEG  [9, 9, 8, 7, 8, 9, 8, 9, 9, 7, 8, 8, 8, 8, 8, ...   \n",
 815 |        "ILSVRC2012_val_00006124.JPEG  [8, 8, 8, 9, 9, 8, 9, 9, 9, 7, 8, 8, 9, 9, 9, ...   \n",
 816 |        "\n",
 817 |        "                              image_number  imagenet_label  num_queries  \\\n",
 818 |        "image                                                                     \n",
 819 |        "ILSVRC2012_val_00047683.JPEG         16296             325           12   \n",
 820 |        "ILSVRC2012_val_00030990.JPEG          2382              47           11   \n",
 821 |        "ILSVRC2012_val_00002431.JPEG         35201             704           17   \n",
 822 |        "ILSVRC2012_val_00033061.JPEG         22133             442           11   \n",
 823 |        "ILSVRC2012_val_00002735.JPEG         33403             668            9   \n",
 824 |        "...                                    ...             ...          ...   \n",
 825 |        "ILSVRC2012_val_00003714.JPEG         10702             214           19   \n",
 826 |        "ILSVRC2012_val_00018015.JPEG         33564             671           23   \n",
 827 |        "ILSVRC2012_val_00029538.JPEG         20129             402           13   \n",
 828 |        "ILSVRC2012_val_00040262.JPEG         46093             921           19   \n",
 829 |        "ILSVRC2012_val_00006124.JPEG         36454             729           18   \n",
 830 |        "\n",
 831 |        "                                                                       sel_dict  \n",
 832 |        "image                                                                            \n",
 833 |        "ILSVRC2012_val_00047683.JPEG  {392: (0, 9), 310: (0, 9), 322: (4, 9), 712: (...  \n",
 834 |        "ILSVRC2012_val_00030990.JPEG  {312: (0, 7), 30: (0, 9), 39: (8, 8), 31: (0, ...  \n",
 835 |        "ILSVRC2012_val_00002431.JPEG  {870: (1, 9), 444: (0, 9), 670: (0, 8), 826: (...  \n",
 836 |        "ILSVRC2012_val_00033061.JPEG  {497: (6, 6), 442: (8, 8), 448: (0, 9), 494: (...  \n",
 837 |        "ILSVRC2012_val_00002735.JPEG  {497: (6, 8), 442: (3, 8), 884: (4, 8), 698: (...  \n",
 838 |        "...                                                                         ...  \n",
 839 |        "ILSVRC2012_val_00003714.JPEG  {214: (3, 9), 177: (1, 6), 170: (2, 8), 231: (...  \n",
 840 |        "ILSVRC2012_val_00018015.JPEG  {843: (0, 9), 671: (2, 9), 892: (1, 9), 871: (...  \n",
 841 |        "ILSVRC2012_val_00029538.JPEG  {610: (0, 9), 714: (0, 9), 566: (0, 9), 745: (...  \n",
 842 |        "ILSVRC2012_val_00040262.JPEG  {794: (0, 9), 440: (0, 9), 868: (0, 8), 585: (...  \n",
 843 |        "ILSVRC2012_val_00006124.JPEG  {758: (1, 8), 726: (1, 8), 589: (0, 8), 618: (...  \n",
 844 |        "\n",
 845 |        "[10000 rows x 8 columns]"
 846 |       ]
 847 |      },
 848 |      "execution_count": 7,
 849 |      "metadata": {},
 850 |      "output_type": "execute_result"
 851 |     }
 852 |    ],
 853 |    "source": [
 854 |     "info_contains"
 855 |    ]
 856 |   },
 857 |   {
 858 |    "cell_type": "markdown",
 859 |    "metadata": {},
 860 |    "source": [
 861 |     "## Split of 10k images used for CONTAINS task\n",
 862 |     "\n",
 863 |     "____________________________________________________________________________________________________\n",
 864 |     "\n",
 865 |     "`image`: image filename \n",
 866 |     "\n",
 867 |     "`annotation`: Can take one of the following 4 values\n",
 868 |     "\n",
 869 |     "(a) `easy`: ImageNet label was the only label selected confidently in the ``valid labels`` expt (no other label has even half ImageNet label sf as per at least 5 workers)\n",
 870 |     "\n",
 871 |     "(b) `mis`: ImageNet label was not selected by any annotators\n",
 872 |     "\n",
 873 |     "(c) `fu`: Images were part of the follow-up experiment due to multiple confident labels\n",
 874 |     "\n",
 875 |     "(a) `amb`: Images omitted from follow-up because number of workers was too low (HITs discarded during controls)"
 876 |    ]
 877 |   },
 878 |   {
 879 |    "cell_type": "code",
 880 |    "execution_count": 8,
 881 |    "metadata": {},
 882 |    "outputs": [],
 883 |    "source": [
 884 |     "info_part = pd.read_pickle(\"./data/image_info.pkl\")"
 885 |    ]
 886 |   },
 887 |   {
 888 |    "cell_type": "code",
 889 |    "execution_count": 9,
 890 |    "metadata": {},
 891 |    "outputs": [
 892 |     {
 893 |      "data": {
 894 |       "text/html": [
 895 |        "<div>\n",
 896 |        "<style scoped>\n",
 897 |        "    .dataframe tbody tr th:only-of-type {\n",
 898 |        "        vertical-align: middle;\n",
 899 |        "    }\n",
 900 |        "\n",
 901 |        "    .dataframe tbody tr th {\n",
 902 |        "        vertical-align: top;\n",
 903 |        "    }\n",
 904 |        "\n",
 905 |        "    .dataframe thead th {\n",
 906 |        "        text-align: right;\n",
 907 |        "    }\n",
 908 |        "</style>\n",
 909 |        "<table border=\"1\" class=\"dataframe\">\n",
 910 |        "  <thead>\n",
 911 |        "    <tr style=\"text-align: right;\">\n",
 912 |        "      <th></th>\n",
 913 |        "      <th>annotation</th>\n",
 914 |        "    </tr>\n",
 915 |        "    <tr>\n",
 916 |        "      <th>image</th>\n",
 917 |        "      <th></th>\n",
 918 |        "    </tr>\n",
 919 |        "  </thead>\n",
 920 |        "  <tbody>\n",
 921 |        "    <tr>\n",
 922 |        "      <th>ILSVRC2012_val_00047683.JPEG</th>\n",
 923 |        "      <td>fu</td>\n",
 924 |        "    </tr>\n",
 925 |        "    <tr>\n",
 926 |        "      <th>ILSVRC2012_val_00030990.JPEG</th>\n",
 927 |        "      <td>fu</td>\n",
 928 |        "    </tr>\n",
 929 |        "    <tr>\n",
 930 |        "      <th>ILSVRC2012_val_00002431.JPEG</th>\n",
 931 |        "      <td>easy</td>\n",
 932 |        "    </tr>\n",
 933 |        "    <tr>\n",
 934 |        "      <th>ILSVRC2012_val_00033061.JPEG</th>\n",
 935 |        "      <td>fu</td>\n",
 936 |        "    </tr>\n",
 937 |        "    <tr>\n",
 938 |        "      <th>ILSVRC2012_val_00002735.JPEG</th>\n",
 939 |        "      <td>fu</td>\n",
 940 |        "    </tr>\n",
 941 |        "    <tr>\n",
 942 |        "      <th>...</th>\n",
 943 |        "      <td>...</td>\n",
 944 |        "    </tr>\n",
 945 |        "    <tr>\n",
 946 |        "      <th>ILSVRC2012_val_00003714.JPEG</th>\n",
 947 |        "      <td>fu</td>\n",
 948 |        "    </tr>\n",
 949 |        "    <tr>\n",
 950 |        "      <th>ILSVRC2012_val_00018015.JPEG</th>\n",
 951 |        "      <td>fu</td>\n",
 952 |        "    </tr>\n",
 953 |        "    <tr>\n",
 954 |        "      <th>ILSVRC2012_val_00029538.JPEG</th>\n",
 955 |        "      <td>easy</td>\n",
 956 |        "    </tr>\n",
 957 |        "    <tr>\n",
 958 |        "      <th>ILSVRC2012_val_00040262.JPEG</th>\n",
 959 |        "      <td>easy</td>\n",
 960 |        "    </tr>\n",
 961 |        "    <tr>\n",
 962 |        "      <th>ILSVRC2012_val_00006124.JPEG</th>\n",
 963 |        "      <td>fu</td>\n",
 964 |        "    </tr>\n",
 965 |        "  </tbody>\n",
 966 |        "</table>\n",
 967 |        "<p>10000 rows × 1 columns</p>\n",
 968 |        "</div>"
 969 |       ],
 970 |       "text/plain": [
 971 |        "                             annotation\n",
 972 |        "image                                  \n",
 973 |        "ILSVRC2012_val_00047683.JPEG         fu\n",
 974 |        "ILSVRC2012_val_00030990.JPEG         fu\n",
 975 |        "ILSVRC2012_val_00002431.JPEG       easy\n",
 976 |        "ILSVRC2012_val_00033061.JPEG         fu\n",
 977 |        "ILSVRC2012_val_00002735.JPEG         fu\n",
 978 |        "...                                 ...\n",
 979 |        "ILSVRC2012_val_00003714.JPEG         fu\n",
 980 |        "ILSVRC2012_val_00018015.JPEG         fu\n",
 981 |        "ILSVRC2012_val_00029538.JPEG       easy\n",
 982 |        "ILSVRC2012_val_00040262.JPEG       easy\n",
 983 |        "ILSVRC2012_val_00006124.JPEG         fu\n",
 984 |        "\n",
 985 |        "[10000 rows x 1 columns]"
 986 |       ]
 987 |      },
 988 |      "execution_count": 9,
 989 |      "metadata": {},
 990 |      "output_type": "execute_result"
 991 |     }
 992 |    ],
 993 |    "source": [
 994 |     "info_part"
 995 |    ]
 996 |   },
 997 |   {
 998 |    "cell_type": "markdown",
 999 |    "metadata": {},
1000 |    "source": [
1001 |     "## Image Classification (CLASSIFY task)\n",
1002 |     "\n",
1003 |     "The dataframe contains the following information:\n",
1004 |     "\n",
1005 |     "____________________________________________________________________________________________________\n",
1006 |     "\n",
1007 |     "`image`: image filename \n",
1008 |     "\n",
1009 |     "`image_number`: number of image as per an unshuffled loader (same as above)\n",
1010 |     "\n",
1011 |     "`imagenet_label`: ImageNet label\n",
1012 |     "\n",
1013 |     "____________________________________________________________________________________________________\n",
1014 |     "\n",
1015 |     "`main`: class for main object in the image, 1 per annotator\n",
1016 |     "\n",
1017 |     "`main_top`: ``main label`` (most frequent entry of ``main``)\n",
1018 |     "\n",
1019 |     "`main_counter`: counter with distribution over label of main object (based on ``main``)\n",
1020 |     "\n",
1021 |     "____________________________________________________________________________________________________\n",
1022 |     "\n",
1023 |     "`selected`: set of labels (corresponding to distinct objects) selected by each annotator\n",
1024 |     "\n",
1025 |     "`num_objects`: number of objects in image (argmax over length of `selected`)\n",
1026 |     "\n",
1027 |     "`object_counter`: distribution over number of objects in image (distribution over length of `selected`)\n",
1028 |     "\n",
1029 |     "`objects`: list of labels for each of the objects in image (total=`num_objects`) obtained by applying max-k cut on `selected`\n",
1030 |     "\n",
1031 |     "`main_object`: object containing `main_top`\n",
1032 |     "\n",
1033 |     "`main_object_counter`: how likely each object is to be `main`\n",
1034 |     "\n",
1035 |     "`imagenet_object`: object corresponding to ImageNet label"
1036 |    ]
1037 |   },
1038 |   {
1039 |    "cell_type": "code",
1040 |    "execution_count": 10,
1041 |    "metadata": {},
1042 |    "outputs": [],
1043 |    "source": [
1044 |     "info_classify = pd.read_pickle(\"./data/annotations_classify_task.pkl\")"
1045 |    ]
1046 |   },
1047 |   {
1048 |    "cell_type": "code",
1049 |    "execution_count": 11,
1050 |    "metadata": {},
1051 |    "outputs": [
1052 |     {
1053 |      "data": {
1054 |       "text/html": [
1055 |        "<div>\n",
1056 |        "<style scoped>\n",
1057 |        "    .dataframe tbody tr th:only-of-type {\n",
1058 |        "        vertical-align: middle;\n",
1059 |        "    }\n",
1060 |        "\n",
1061 |        "    .dataframe tbody tr th {\n",
1062 |        "        vertical-align: top;\n",
1063 |        "    }\n",
1064 |        "\n",
1065 |        "    .dataframe thead th {\n",
1066 |        "        text-align: right;\n",
1067 |        "    }\n",
1068 |        "</style>\n",
1069 |        "<table border=\"1\" class=\"dataframe\">\n",
1070 |        "  <thead>\n",
1071 |        "    <tr style=\"text-align: right;\">\n",
1072 |        "      <th></th>\n",
1073 |        "      <th>main</th>\n",
1074 |        "      <th>selected</th>\n",
1075 |        "      <th>num_workers</th>\n",
1076 |        "      <th>image_number</th>\n",
1077 |        "      <th>imagenet_label</th>\n",
1078 |        "      <th>main_counter</th>\n",
1079 |        "      <th>main_top</th>\n",
1080 |        "      <th>num_objects</th>\n",
1081 |        "      <th>object_counter</th>\n",
1082 |        "      <th>objects</th>\n",
1083 |        "      <th>main_object</th>\n",
1084 |        "      <th>main_object_counter</th>\n",
1085 |        "      <th>imagenet_object</th>\n",
1086 |        "    </tr>\n",
1087 |        "    <tr>\n",
1088 |        "      <th>image</th>\n",
1089 |        "      <th></th>\n",
1090 |        "      <th></th>\n",
1091 |        "      <th></th>\n",
1092 |        "      <th></th>\n",
1093 |        "      <th></th>\n",
1094 |        "      <th></th>\n",
1095 |        "      <th></th>\n",
1096 |        "      <th></th>\n",
1097 |        "      <th></th>\n",
1098 |        "      <th></th>\n",
1099 |        "      <th></th>\n",
1100 |        "      <th></th>\n",
1101 |        "      <th></th>\n",
1102 |        "    </tr>\n",
1103 |        "  </thead>\n",
1104 |        "  <tbody>\n",
1105 |        "    <tr>\n",
1106 |        "      <th>ILSVRC2012_val_00047683.JPEG</th>\n",
1107 |        "      <td>[325, 322, 322, 325, 322, 322, 325, 324, 322]</td>\n",
1108 |        "      <td>[[326, 322, 325], [322], [322], [325], [322], ...</td>\n",
1109 |        "      <td>9</td>\n",
1110 |        "      <td>16296</td>\n",
1111 |        "      <td>325</td>\n",
1112 |        "      <td>{325: 3, 322: 5, 324: 1}</td>\n",
1113 |        "      <td>322</td>\n",
1114 |        "      <td>1</td>\n",
1115 |        "      <td>{3: 1, 1: 8}</td>\n",
1116 |        "      <td>[{326: (0.09090909090909091, 0.111111111111111...</td>\n",
1117 |        "      <td>0</td>\n",
1118 |        "      <td>[9]</td>\n",
1119 |        "      <td>0.0</td>\n",
1120 |        "    </tr>\n",
1121 |        "    <tr>\n",
1122 |        "      <th>ILSVRC2012_val_00030990.JPEG</th>\n",
1123 |        "      <td>[40, 47, 39, 47, 47, 40, 47, 40, 47]</td>\n",
1124 |        "      <td>[[40], [47], [39], [47], [47], [40], [47], [40...</td>\n",
1125 |        "      <td>9</td>\n",
1126 |        "      <td>2382</td>\n",
1127 |        "      <td>47</td>\n",
1128 |        "      <td>{40: 3, 47: 5, 39: 1}</td>\n",
1129 |        "      <td>47</td>\n",
1130 |        "      <td>1</td>\n",
1131 |        "      <td>{1: 7, 2: 2}</td>\n",
1132 |        "      <td>[{40: (0.36363636363636365, 0.4444444444444444...</td>\n",
1133 |        "      <td>0</td>\n",
1134 |        "      <td>[9]</td>\n",
1135 |        "      <td>0.0</td>\n",
1136 |        "    </tr>\n",
1137 |        "    <tr>\n",
1138 |        "      <th>ILSVRC2012_val_00033061.JPEG</th>\n",
1139 |        "      <td>[497, 663, 497, 442, 497, 497, 497, 497, 497]</td>\n",
1140 |        "      <td>[[497, 442], [497, 442, 663], [497, 442], [442...</td>\n",
1141 |        "      <td>9</td>\n",
1142 |        "      <td>22133</td>\n",
1143 |        "      <td>442</td>\n",
1144 |        "      <td>{497: 7, 663: 1, 442: 1}</td>\n",
1145 |        "      <td>497</td>\n",
1146 |        "      <td>2</td>\n",
1147 |        "      <td>{2: 3, 3: 3, 1: 3}</td>\n",
1148 |        "      <td>[{442: (1.0, 0.7777777777777778)}, {497: (0.72...</td>\n",
1149 |        "      <td>1</td>\n",
1150 |        "      <td>[1, 8]</td>\n",
1151 |        "      <td>0.0</td>\n",
1152 |        "    </tr>\n",
1153 |        "    <tr>\n",
1154 |        "      <th>ILSVRC2012_val_00002735.JPEG</th>\n",
1155 |        "      <td>[497, 832, 663, 668, 832, 668, 538, 668, 668]</td>\n",
1156 |        "      <td>[[538, 497], [832, 538, 668, 663, 884, 497], [...</td>\n",
1157 |        "      <td>9</td>\n",
1158 |        "      <td>33403</td>\n",
1159 |        "      <td>668</td>\n",
1160 |        "      <td>{497: 1, 832: 2, 663: 1, 668: 4, 538: 1}</td>\n",
1161 |        "      <td>668</td>\n",
1162 |        "      <td>2</td>\n",
1163 |        "      <td>{2: 3, 6: 1, 4: 3, 1: 2}</td>\n",
1164 |        "      <td>[{884: (0.15384615384615385, 0.222222222222222...</td>\n",
1165 |        "      <td>1</td>\n",
1166 |        "      <td>[2, 7]</td>\n",
1167 |        "      <td>1.0</td>\n",
1168 |        "    </tr>\n",
1169 |        "    <tr>\n",
1170 |        "      <th>ILSVRC2012_val_00019033.JPEG</th>\n",
1171 |        "      <td>[69, 69, 69, 69, 116, 116, 69, 116, 126]</td>\n",
1172 |        "      <td>[[69], [69], [69], [69], [116], [69, 126, 116]...</td>\n",
1173 |        "      <td>9</td>\n",
1174 |        "      <td>5818</td>\n",
1175 |        "      <td>116</td>\n",
1176 |        "      <td>{69: 5, 116: 3, 126: 1}</td>\n",
1177 |        "      <td>69</td>\n",
1178 |        "      <td>1</td>\n",
1179 |        "      <td>{1: 7, 3: 2}</td>\n",
1180 |        "      <td>[{69: (0.46153846153846156, 0.6666666666666666...</td>\n",
1181 |        "      <td>0</td>\n",
1182 |        "      <td>[9]</td>\n",
1183 |        "      <td>0.0</td>\n",
1184 |        "    </tr>\n",
1185 |        "    <tr>\n",
1186 |        "      <th>...</th>\n",
1187 |        "      <td>...</td>\n",
1188 |        "      <td>...</td>\n",
1189 |        "      <td>...</td>\n",
1190 |        "      <td>...</td>\n",
1191 |        "      <td>...</td>\n",
1192 |        "      <td>...</td>\n",
1193 |        "      <td>...</td>\n",
1194 |        "      <td>...</td>\n",
1195 |        "      <td>...</td>\n",
1196 |        "      <td>...</td>\n",
1197 |        "      <td>...</td>\n",
1198 |        "      <td>...</td>\n",
1199 |        "      <td>...</td>\n",
1200 |        "    </tr>\n",
1201 |        "    <tr>\n",
1202 |        "      <th>ILSVRC2012_val_00025813.JPEG</th>\n",
1203 |        "      <td>[947, 947, 947, 947, 947, 947, 947, 947, 947]</td>\n",
1204 |        "      <td>[[997, 947], [997, 947], [947], [947], [997, 9...</td>\n",
1205 |        "      <td>9</td>\n",
1206 |        "      <td>47377</td>\n",
1207 |        "      <td>947</td>\n",
1208 |        "      <td>{947: 9}</td>\n",
1209 |        "      <td>947</td>\n",
1210 |        "      <td>1</td>\n",
1211 |        "      <td>{2: 4, 1: 5}</td>\n",
1212 |        "      <td>[{997: (0.3076923076923077, 0.4444444444444444...</td>\n",
1213 |        "      <td>0</td>\n",
1214 |        "      <td>[9]</td>\n",
1215 |        "      <td>0.0</td>\n",
1216 |        "    </tr>\n",
1217 |        "    <tr>\n",
1218 |        "      <th>ILSVRC2012_val_00040269.JPEG</th>\n",
1219 |        "      <td>[382, 381, 381, 382, 382, 382, 382, 382, 382]</td>\n",
1220 |        "      <td>[[382], [381], [381], [382], [382], [382], [38...</td>\n",
1221 |        "      <td>9</td>\n",
1222 |        "      <td>19041</td>\n",
1223 |        "      <td>380</td>\n",
1224 |        "      <td>{382: 7, 381: 2}</td>\n",
1225 |        "      <td>382</td>\n",
1226 |        "      <td>1</td>\n",
1227 |        "      <td>{1: 8, 2: 1}</td>\n",
1228 |        "      <td>[{382: (0.7, 0.7777777777777778), 381: (0.3, 0...</td>\n",
1229 |        "      <td>0</td>\n",
1230 |        "      <td>[9]</td>\n",
1231 |        "      <td>NaN</td>\n",
1232 |        "    </tr>\n",
1233 |        "    <tr>\n",
1234 |        "      <th>ILSVRC2012_val_00003714.JPEG</th>\n",
1235 |        "      <td>[214, 256, 240, 214, 256, 214, 205, 214, 214]</td>\n",
1236 |        "      <td>[[214, 240, 256, 236], [256], [240], [214, 256...</td>\n",
1237 |        "      <td>9</td>\n",
1238 |        "      <td>10702</td>\n",
1239 |        "      <td>214</td>\n",
1240 |        "      <td>{214: 5, 256: 2, 240: 1, 205: 1}</td>\n",
1241 |        "      <td>214</td>\n",
1242 |        "      <td>1</td>\n",
1243 |        "      <td>{4: 1, 1: 7, 2: 1}</td>\n",
1244 |        "      <td>[{214: (0.38461538461538464, 0.555555555555555...</td>\n",
1245 |        "      <td>0</td>\n",
1246 |        "      <td>[9]</td>\n",
1247 |        "      <td>0.0</td>\n",
1248 |        "    </tr>\n",
1249 |        "    <tr>\n",
1250 |        "      <th>ILSVRC2012_val_00018015.JPEG</th>\n",
1251 |        "      <td>[671, 671, 671, 671, 671, 671, 671, 671, 671]</td>\n",
1252 |        "      <td>[[535, 671], [671], [671], [671], [671], [671]...</td>\n",
1253 |        "      <td>9</td>\n",
1254 |        "      <td>33564</td>\n",
1255 |        "      <td>671</td>\n",
1256 |        "      <td>{671: 9}</td>\n",
1257 |        "      <td>671</td>\n",
1258 |        "      <td>1</td>\n",
1259 |        "      <td>{2: 1, 1: 8}</td>\n",
1260 |        "      <td>[{535: (0.1, 0.1111111111111111), 671: (0.9, 1...</td>\n",
1261 |        "      <td>0</td>\n",
1262 |        "      <td>[9]</td>\n",
1263 |        "      <td>0.0</td>\n",
1264 |        "    </tr>\n",
1265 |        "    <tr>\n",
1266 |        "      <th>ILSVRC2012_val_00006124.JPEG</th>\n",
1267 |        "      <td>[729, 729, 729, 729, 729, 729, 729, 729, 729]</td>\n",
1268 |        "      <td>[[828, 729, 567], [828, 729], [828, 729], [828...</td>\n",
1269 |        "      <td>9</td>\n",
1270 |        "      <td>36454</td>\n",
1271 |        "      <td>729</td>\n",
1272 |        "      <td>{729: 9}</td>\n",
1273 |        "      <td>729</td>\n",
1274 |        "      <td>2</td>\n",
1275 |        "      <td>{3: 2, 2: 6, 1: 1}</td>\n",
1276 |        "      <td>[{828: (1.0, 0.8888888888888888)}, {729: (0.81...</td>\n",
1277 |        "      <td>1</td>\n",
1278 |        "      <td>[0, 9]</td>\n",
1279 |        "      <td>1.0</td>\n",
1280 |        "    </tr>\n",
1281 |        "  </tbody>\n",
1282 |        "</table>\n",
1283 |        "<p>6761 rows × 13 columns</p>\n",
1284 |        "</div>"
1285 |       ],
1286 |       "text/plain": [
1287 |        "                                                                       main  \\\n",
1288 |        "image                                                                         \n",
1289 |        "ILSVRC2012_val_00047683.JPEG  [325, 322, 322, 325, 322, 322, 325, 324, 322]   \n",
1290 |        "ILSVRC2012_val_00030990.JPEG           [40, 47, 39, 47, 47, 40, 47, 40, 47]   \n",
1291 |        "ILSVRC2012_val_00033061.JPEG  [497, 663, 497, 442, 497, 497, 497, 497, 497]   \n",
1292 |        "ILSVRC2012_val_00002735.JPEG  [497, 832, 663, 668, 832, 668, 538, 668, 668]   \n",
1293 |        "ILSVRC2012_val_00019033.JPEG       [69, 69, 69, 69, 116, 116, 69, 116, 126]   \n",
1294 |        "...                                                                     ...   \n",
1295 |        "ILSVRC2012_val_00025813.JPEG  [947, 947, 947, 947, 947, 947, 947, 947, 947]   \n",
1296 |        "ILSVRC2012_val_00040269.JPEG  [382, 381, 381, 382, 382, 382, 382, 382, 382]   \n",
1297 |        "ILSVRC2012_val_00003714.JPEG  [214, 256, 240, 214, 256, 214, 205, 214, 214]   \n",
1298 |        "ILSVRC2012_val_00018015.JPEG  [671, 671, 671, 671, 671, 671, 671, 671, 671]   \n",
1299 |        "ILSVRC2012_val_00006124.JPEG  [729, 729, 729, 729, 729, 729, 729, 729, 729]   \n",
1300 |        "\n",
1301 |        "                                                                       selected  \\\n",
1302 |        "image                                                                             \n",
1303 |        "ILSVRC2012_val_00047683.JPEG  [[326, 322, 325], [322], [322], [325], [322], ...   \n",
1304 |        "ILSVRC2012_val_00030990.JPEG  [[40], [47], [39], [47], [47], [40], [47], [40...   \n",
1305 |        "ILSVRC2012_val_00033061.JPEG  [[497, 442], [497, 442, 663], [497, 442], [442...   \n",
1306 |        "ILSVRC2012_val_00002735.JPEG  [[538, 497], [832, 538, 668, 663, 884, 497], [...   \n",
1307 |        "ILSVRC2012_val_00019033.JPEG  [[69], [69], [69], [69], [116], [69, 126, 116]...   \n",
1308 |        "...                                                                         ...   \n",
1309 |        "ILSVRC2012_val_00025813.JPEG  [[997, 947], [997, 947], [947], [947], [997, 9...   \n",
1310 |        "ILSVRC2012_val_00040269.JPEG  [[382], [381], [381], [382], [382], [382], [38...   \n",
1311 |        "ILSVRC2012_val_00003714.JPEG  [[214, 240, 256, 236], [256], [240], [214, 256...   \n",
1312 |        "ILSVRC2012_val_00018015.JPEG  [[535, 671], [671], [671], [671], [671], [671]...   \n",
1313 |        "ILSVRC2012_val_00006124.JPEG  [[828, 729, 567], [828, 729], [828, 729], [828...   \n",
1314 |        "\n",
1315 |        "                              num_workers  image_number  imagenet_label  \\\n",
1316 |        "image                                                                     \n",
1317 |        "ILSVRC2012_val_00047683.JPEG            9         16296             325   \n",
1318 |        "ILSVRC2012_val_00030990.JPEG            9          2382              47   \n",
1319 |        "ILSVRC2012_val_00033061.JPEG            9         22133             442   \n",
1320 |        "ILSVRC2012_val_00002735.JPEG            9         33403             668   \n",
1321 |        "ILSVRC2012_val_00019033.JPEG            9          5818             116   \n",
1322 |        "...                                   ...           ...             ...   \n",
1323 |        "ILSVRC2012_val_00025813.JPEG            9         47377             947   \n",
1324 |        "ILSVRC2012_val_00040269.JPEG            9         19041             380   \n",
1325 |        "ILSVRC2012_val_00003714.JPEG            9         10702             214   \n",
1326 |        "ILSVRC2012_val_00018015.JPEG            9         33564             671   \n",
1327 |        "ILSVRC2012_val_00006124.JPEG            9         36454             729   \n",
1328 |        "\n",
1329 |        "                                                          main_counter  \\\n",
1330 |        "image                                                                    \n",
1331 |        "ILSVRC2012_val_00047683.JPEG                  {325: 3, 322: 5, 324: 1}   \n",
1332 |        "ILSVRC2012_val_00030990.JPEG                     {40: 3, 47: 5, 39: 1}   \n",
1333 |        "ILSVRC2012_val_00033061.JPEG                  {497: 7, 663: 1, 442: 1}   \n",
1334 |        "ILSVRC2012_val_00002735.JPEG  {497: 1, 832: 2, 663: 1, 668: 4, 538: 1}   \n",
1335 |        "ILSVRC2012_val_00019033.JPEG                   {69: 5, 116: 3, 126: 1}   \n",
1336 |        "...                                                                ...   \n",
1337 |        "ILSVRC2012_val_00025813.JPEG                                  {947: 9}   \n",
1338 |        "ILSVRC2012_val_00040269.JPEG                          {382: 7, 381: 2}   \n",
1339 |        "ILSVRC2012_val_00003714.JPEG          {214: 5, 256: 2, 240: 1, 205: 1}   \n",
1340 |        "ILSVRC2012_val_00018015.JPEG                                  {671: 9}   \n",
1341 |        "ILSVRC2012_val_00006124.JPEG                                  {729: 9}   \n",
1342 |        "\n",
1343 |        "                              main_top  num_objects            object_counter  \\\n",
1344 |        "image                                                                           \n",
1345 |        "ILSVRC2012_val_00047683.JPEG       322            1              {3: 1, 1: 8}   \n",
1346 |        "ILSVRC2012_val_00030990.JPEG        47            1              {1: 7, 2: 2}   \n",
1347 |        "ILSVRC2012_val_00033061.JPEG       497            2        {2: 3, 3: 3, 1: 3}   \n",
1348 |        "ILSVRC2012_val_00002735.JPEG       668            2  {2: 3, 6: 1, 4: 3, 1: 2}   \n",
1349 |        "ILSVRC2012_val_00019033.JPEG        69            1              {1: 7, 3: 2}   \n",
1350 |        "...                                ...          ...                       ...   \n",
1351 |        "ILSVRC2012_val_00025813.JPEG       947            1              {2: 4, 1: 5}   \n",
1352 |        "ILSVRC2012_val_00040269.JPEG       382            1              {1: 8, 2: 1}   \n",
1353 |        "ILSVRC2012_val_00003714.JPEG       214            1        {4: 1, 1: 7, 2: 1}   \n",
1354 |        "ILSVRC2012_val_00018015.JPEG       671            1              {2: 1, 1: 8}   \n",
1355 |        "ILSVRC2012_val_00006124.JPEG       729            2        {3: 2, 2: 6, 1: 1}   \n",
1356 |        "\n",
1357 |        "                                                                        objects  \\\n",
1358 |        "image                                                                             \n",
1359 |        "ILSVRC2012_val_00047683.JPEG  [{326: (0.09090909090909091, 0.111111111111111...   \n",
1360 |        "ILSVRC2012_val_00030990.JPEG  [{40: (0.36363636363636365, 0.4444444444444444...   \n",
1361 |        "ILSVRC2012_val_00033061.JPEG  [{442: (1.0, 0.7777777777777778)}, {497: (0.72...   \n",
1362 |        "ILSVRC2012_val_00002735.JPEG  [{884: (0.15384615384615385, 0.222222222222222...   \n",
1363 |        "ILSVRC2012_val_00019033.JPEG  [{69: (0.46153846153846156, 0.6666666666666666...   \n",
1364 |        "...                                                                         ...   \n",
1365 |        "ILSVRC2012_val_00025813.JPEG  [{997: (0.3076923076923077, 0.4444444444444444...   \n",
1366 |        "ILSVRC2012_val_00040269.JPEG  [{382: (0.7, 0.7777777777777778), 381: (0.3, 0...   \n",
1367 |        "ILSVRC2012_val_00003714.JPEG  [{214: (0.38461538461538464, 0.555555555555555...   \n",
1368 |        "ILSVRC2012_val_00018015.JPEG  [{535: (0.1, 0.1111111111111111), 671: (0.9, 1...   \n",
1369 |        "ILSVRC2012_val_00006124.JPEG  [{828: (1.0, 0.8888888888888888)}, {729: (0.81...   \n",
1370 |        "\n",
1371 |        "                              main_object main_object_counter  imagenet_object  \n",
1372 |        "image                                                                           \n",
1373 |        "ILSVRC2012_val_00047683.JPEG            0                 [9]              0.0  \n",
1374 |        "ILSVRC2012_val_00030990.JPEG            0                 [9]              0.0  \n",
1375 |        "ILSVRC2012_val_00033061.JPEG            1              [1, 8]              0.0  \n",
1376 |        "ILSVRC2012_val_00002735.JPEG            1              [2, 7]              1.0  \n",
1377 |        "ILSVRC2012_val_00019033.JPEG            0                 [9]              0.0  \n",
1378 |        "...                                   ...                 ...              ...  \n",
1379 |        "ILSVRC2012_val_00025813.JPEG            0                 [9]              0.0  \n",
1380 |        "ILSVRC2012_val_00040269.JPEG            0                 [9]              NaN  \n",
1381 |        "ILSVRC2012_val_00003714.JPEG            0                 [9]              0.0  \n",
1382 |        "ILSVRC2012_val_00018015.JPEG            0                 [9]              0.0  \n",
1383 |        "ILSVRC2012_val_00006124.JPEG            1              [0, 9]              1.0  \n",
1384 |        "\n",
1385 |        "[6761 rows x 13 columns]"
1386 |       ]
1387 |      },
1388 |      "execution_count": 11,
1389 |      "metadata": {},
1390 |      "output_type": "execute_result"
1391 |     }
1392 |    ],
1393 |    "source": [
1394 |     "info_classify"
1395 |    ]
1396 |   },
1397 |   {
1398 |    "cell_type": "markdown",
1399 |    "metadata": {},
1400 |    "source": [
1401 |     "### Visualize fine-grained image annotations"
1402 |    ]
1403 |   },
1404 |   {
1405 |    "cell_type": "code",
1406 |    "execution_count": 12,
1407 |    "metadata": {},
1408 |    "outputs": [
1409 |     {
1410 |      "data": {
1411 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAJ8AAADnCAYAAAAAaoRMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
1412 |       "text/plain": [
1413 |        "<Figure size 432x288 with 1 Axes>"
1414 |       ]
1415 |      },
1416 |      "metadata": {},
1417 |      "output_type": "display_data"
1418 |     },
1419 |     {
1420 |      "name": "stdout",
1421 |      "output_type": "stream",
1422 |      "text": [
1423 |       "ImageNet label: bell cote, bell cot  \n",
1424 |       "\n",
1425 |       "---------------\n",
1426 |       "\n",
1427 |       "# Objects: 2\n",
1428 |       "Main label: church, church building\n",
1429 |       "\n",
1430 |       "Objects:\n",
1431 |       "church, church building\n",
1432 |       "bell cote, bell cot\n"
1433 |      ]
1434 |     }
1435 |    ],
1436 |    "source": [
1437 |     "IMAGENET_PATH = '/scratch/datasets/imagenet-pytorch/val'\n",
1438 |     "\n",
1439 |     "vis_idx = 2\n",
1440 |     "\n",
1441 |     "img, label = info_classify.index[vis_idx], info_classify.iloc[vis_idx]['imagenet_label']\n",
1442 |     "image_path = f\"{IMAGENET_PATH}/{label_to_wnid[label]}/{img}\"\n",
1443 |     "annotations = {k: info_classify.iloc[vis_idx][k] for k in ['main_top', 'num_objects', 'objects']}\n",
1444 |     "objects = set([sorted((-v[0], k) for k, v in o.items())[0][1] for o in annotations['objects']])\n",
1445 |     "\n",
1446 |     "img = Image.open(image_path)\n",
1447 |     "plt.imshow(img)\n",
1448 |     "plt.axis('off')\n",
1449 |     "plt.show()\n",
1450 |     "\n",
1451 |     "\n",
1452 |     "print(f\"ImageNet label: {label_map[label]}  \\n\")\n",
1453 |     "print(f\"---------------\\n\")\n",
1454 |     "print(f\"# Objects: {annotations['num_objects']}\")\n",
1455 |     "print(f\"Main label: {label_map[annotations['main_top']]}\\n\")\n",
1456 |     "print(\"Objects:\")\n",
1457 |     "for oi, o in enumerate(objects):\n",
1458 |     "    print(f\"{label_map[o]}\")\n"
1459 |    ]
1460 |   },
1461 |   {
1462 |    "cell_type": "code",
1463 |    "execution_count": null,
1464 |    "metadata": {},
1465 |    "outputs": [],
1466 |    "source": []
1467 |   }
1468 |  ],
1469 |  "metadata": {
1470 |   "kernelspec": {
1471 |    "display_name": "Python 3",
1472 |    "language": "python",
1473 |    "name": "python3"
1474 |   },
1475 |   "language_info": {
1476 |    "codemirror_mode": {
1477 |     "name": "ipython",
1478 |     "version": 3
1479 |    },
1480 |    "file_extension": ".py",
1481 |    "mimetype": "text/x-python",
1482 |    "name": "python",
1483 |    "nbconvert_exporter": "python",
1484 |    "pygments_lexer": "ipython3",
1485 |    "version": "3.6.3"
1486 |   }
1487 |  },
1488 |  "nbformat": 4,
1489 |  "nbformat_minor": 2
1490 | }
1491 | 


--------------------------------------------------------------------------------
/data/annotations_classify_task.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/MadryLab/ImageNetMultiLabel/f939d1f4d4d40ce5e2d77dd82fddd58028781a50/data/annotations_classify_task.pkl


--------------------------------------------------------------------------------
/data/annotations_contains_task.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/MadryLab/ImageNetMultiLabel/f939d1f4d4d40ce5e2d77dd82fddd58028781a50/data/annotations_contains_task.pkl


--------------------------------------------------------------------------------
/data/image_info.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/MadryLab/ImageNetMultiLabel/f939d1f4d4d40ce5e2d77dd82fddd58028781a50/data/image_info.pkl


--------------------------------------------------------------------------------
/data/imagenet_label_map.npy:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/MadryLab/ImageNetMultiLabel/f939d1f4d4d40ce5e2d77dd82fddd58028781a50/data/imagenet_label_map.npy


--------------------------------------------------------------------------------
/data/imagenet_label_to_wnid.npy:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/MadryLab/ImageNetMultiLabel/f939d1f4d4d40ce5e2d77dd82fddd58028781a50/data/imagenet_label_to_wnid.npy


--------------------------------------------------------------------------------
/data/model_predictions.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/MadryLab/ImageNetMultiLabel/f939d1f4d4d40ce5e2d77dd82fddd58028781a50/data/model_predictions.pkl


--------------------------------------------------------------------------------
/data/superclasses.npy:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/MadryLab/ImageNetMultiLabel/f939d1f4d4d40ce5e2d77dd82fddd58028781a50/data/superclasses.npy


--------------------------------------------------------------------------------
/pipeline.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/MadryLab/ImageNetMultiLabel/f939d1f4d4d40ce5e2d77dd82fddd58028781a50/pipeline.jpg


--------------------------------------------------------------------------------