{
  "id": 251079,
  "title": "How to interpret variables?",
  "url": "/competitions/siim-covid19-detection/discussion/251079",
  "author_name": "",
  "post_date": "2021-07-05T19:49:46.620234900Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>How to interpret <strong><em>bounding boxes and labels variables?</em></strong> That said, what is the <strong><em>difference between train image level and train study level files?</em></strong></p>",
  "messages": [
    {
      "id": "1377412",
      "postDate": "07/05/2021 19:49:46",
      "content": "<p>How to interpret <strong><em>bounding boxes and labels variables?</em></strong> That said, what is the <strong><em>difference between train image level and train study level files?</em></strong></p>",
      "rawMarkdown": "How to interpret ***bounding boxes and labels variables?*** That said, what is the ***difference between train image level and train study level files?***",
      "votes": null
    },
    {
      "id": "1377733",
      "postDate": "07/06/2021 04:48:32",
      "content": "<p>The train study level file contains the labels for classification <code>Atypical Appearance, Indeterminate Appearance, Negative for Pneumonia, Typical Appearance</code>.</p>\n<p>The image level file contains bounding box and label data for objection detection.</p>\n<p>Bounding boxes are stored in train_image_level.csv as lists of dicts. Since each image can have more than one BB, the list contains one or more dicts of <code>{x, y, width, height}</code>.</p>\n<p>The labels are a string that contains <code>opacity 1 x y w h</code> .. where the 1 represents the confidence score .. which should always be 1. Each label on an image is concatenated onto one string separated by spaces. </p>\n<p>Multiple labels look like this .. <br>\n<code>opacity 1 789.28836 582.43035 1815.94498 2499.73327 opacity 1 2245.91208 591.20528 3340.5737 2352.75472</code>. </p>\n<p>A label of <code>none 1 0 0 1 1</code> means there is no opacity.</p>",
      "rawMarkdown": "The train study level file contains the labels for classification `Atypical Appearance, Indeterminate Appearance, Negative for Pneumonia, Typical Appearance`.\n\nThe image level file contains bounding box and label data for objection detection.\n\nBounding boxes are stored in train_image_level.csv as lists of dicts. Since each image can have more than one BB, the list contains one or more dicts of `{x, y, width, height}`.\n\nThe labels are a string that contains `opacity 1 x y w h` .. where the 1 represents the confidence score .. which should always be 1. Each label on an image is concatenated onto one string separated by spaces. \n\nMultiple labels look like this .. \n`opacity 1 789.28836 582.43035 1815.94498 2499.73327 opacity 1 2245.91208 591.20528 3340.5737 2352.75472`. \n\nA label of `none 1 0 0 1 1` means there is no opacity.",
      "votes": null
    },
    {
      "id": "1378288",
      "postDate": "07/06/2021 12:31:48",
      "content": "<p>Thank you for elaborating :) What if the confidence is 0? Could you also tell me the best method to plot the bounding boxes on the images please?</p>",
      "rawMarkdown": "Thank you for elaborating :) What if the confidence is 0? Could you also tell me the best method to plot the bounding boxes on the images please?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1377733,
      "author_name": "davidbroberts",
      "author_url": "",
      "post_date": "07/06/2021 04:48:32",
      "content": "<p>The train study level file contains the labels for classification <code>Atypical Appearance, Indeterminate Appearance, Negative for Pneumonia, Typical Appearance</code>.</p>\n<p>The image level file contains bounding box and label data for objection detection.</p>\n<p>Bounding boxes are stored in train_image_level.csv as lists of dicts. Since each image can have more than one BB, the list contains one or more dicts of <code>{x, y, width, height}</code>.</p>\n<p>The labels are a string that contains <code>opacity 1 x y w h</code> .. where the 1 represents the confidence score .. which should always be 1. Each label on an image is concatenated onto one string separated by spaces. </p>\n<p>Multiple labels look like this .. <br>\n<code>opacity 1 789.28836 582.43035 1815.94498 2499.73327 opacity 1 2245.91208 591.20528 3340.5737 2352.75472</code>. </p>\n<p>A label of <code>none 1 0 0 1 1</code> means there is no opacity.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1378288,
          "author_name": "ritikpnayak",
          "author_url": "",
          "post_date": "07/06/2021 12:31:48",
          "content": "<p>Thank you for elaborating :) What if the confidence is 0? Could you also tell me the best method to plot the bounding boxes on the images please?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1377412": "How to interpret ***bounding boxes and labels variables?*** That said, what is the ***difference between train image level and train study level files?***",
    "1377733": "The train study level file contains the labels for classification `Atypical Appearance, Indeterminate Appearance, Negative for Pneumonia, Typical Appearance`.\n\nThe image level file contains bounding box and label data for objection detection.\n\nBounding boxes are stored in train_image_level.csv as lists of dicts. Since each image can have more than one BB, the list contains one or more dicts of `{x, y, width, height}`.\n\nThe labels are a string that contains `opacity 1 x y w h` .. where the 1 represents the confidence score .. which should always be 1. Each label on an image is concatenated onto one string separated by spaces. \n\nMultiple labels look like this .. \n`opacity 1 789.28836 582.43035 1815.94498 2499.73327 opacity 1 2245.91208 591.20528 3340.5737 2352.75472`. \n\nA label of `none 1 0 0 1 1` means there is no opacity.",
    "1378288": "Thank you for elaborating :) What if the confidence is 0? Could you also tell me the best method to plot the bounding boxes on the images please?"
  },
  "source": "meta"
}