{
  "id": 230861,
  "title": "Why are they in folders?",
  "url": "/competitions/herbarium-2021-fgvc8/discussion/230861",
  "author_name": "AndrewTGraham",
  "post_date": "2021-04-05T23:40:39.218000",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Can someone explain the folder system to me. The train images are in numbered folders in numbered folders and the test images are in numbers folders. Why is this? Do the numbers correspond to categories?</p>",
  "messages": [
    {
      "id": 1264494,
      "postDate": "2021-04-06T07:47:41.103Z",
      "content": "<p>Yes, the structure works as follows: </p>\n<p>The training set images are organized in subfolders <code>train/images/&lt;subfolder1&gt;/&lt;subfolder2&gt;/&lt;image id&gt;.jpg</code>, where <code>&lt;subfolder1&gt;</code> combined with <code>&lt;subfolder2&gt;</code> corresponds to the <code>category_id</code>. For example, a training image with an <code>image_id</code> of <code>1104517</code> and a <code>category_id</code> of <code>00001</code>, can be found at <code>train/images/000/01/1104517.jpg</code>. </p>\n<p>The test set images are organized in subfolders <code>test/images/&lt;subfolder&gt;/&lt;image id&gt;.jpg</code>, where <code>&lt;subfolder&gt;</code> corresponds to the integer division of the <code>image_id</code> by 1000. For example, a test image with and <code>image_id</code> of <code>8005</code>, can be found at <code>test/images/008/8005.jpg</code>. </p>",
      "rawMarkdown": "Yes, the structure works as follows: \n\nThe training set images are organized in subfolders `train/images/<subfolder1>/<subfolder2>/<image id>.jpg`, where `<subfolder1>` combined with `<subfolder2>` corresponds to the `category_id`. For example, a training image with an `image_id` of `1104517` and a `category_id` of `00001`, can be found at `train/images/000/01/1104517.jpg`. \n\n\nThe test set images are organized in subfolders `test/images/<subfolder>/<image id>.jpg`, where `<subfolder>` corresponds to the integer division of the `image_id` by 1000. For example, a test image with and `image_id` of `8005`, can be found at `test/images/008/8005.jpg`. ",
      "votes": 3,
      "replies": [
        {
          "id": 1265112,
          "postDate": "2021-04-06T16:21:23.030Z",
          "content": "<p>Thanks a lot! And do the categories correspond to anything of significance, such as species, or are they totally random and just exist to prevent having too many images in one folder which I suppose could be an issue for some computers maybe?</p>\n<p>Additionally, if the categories don't have any significance, are the images in the train folder pre-shuffled?</p>",
          "rawMarkdown": "Thanks a lot! And do the categories correspond to anything of significance, such as species, or are they totally random and just exist to prevent having too many images in one folder which I suppose could be an issue for some computers maybe?\n\nAdditionally, if the categories don't have any significance, are the images in the train folder pre-shuffled?"
        },
        {
          "id": 1265768,
          "postDate": "2021-04-07T07:37:29.623Z",
          "content": "<p>Yes categories correspond to species. For the test set however images are shuffled and in different folders such that no folder has too many files. </p>",
          "rawMarkdown": "Yes categories correspond to species. For the test set however images are shuffled and in different folders such that no folder has too many files. "
        }
      ]
    },
    {
      "id": 1264146,
      "postDate": "2021-04-05T23:40:39.220Z",
      "content": "<p>Can someone explain the folder system to me. The train images are in numbered folders in numbered folders and the test images are in numbers folders. Why is this? Do the numbers correspond to categories?</p>",
      "rawMarkdown": "Can someone explain the folder system to me. The train images are in numbered folders in numbered folders and the test images are in numbers folders. Why is this? Do the numbers correspond to categories?",
      "votes": 1
    }
  ],
  "comments": [
    {
      "id": 1264494,
      "author_name": "Riccardo de Lutio",
      "author_url": "",
      "post_date": "2021-04-06T07:47:41.103000",
      "content": "<p>Yes, the structure works as follows: </p>\n<p>The training set images are organized in subfolders <code>train/images/&lt;subfolder1&gt;/&lt;subfolder2&gt;/&lt;image id&gt;.jpg</code>, where <code>&lt;subfolder1&gt;</code> combined with <code>&lt;subfolder2&gt;</code> corresponds to the <code>category_id</code>. For example, a training image with an <code>image_id</code> of <code>1104517</code> and a <code>category_id</code> of <code>00001</code>, can be found at <code>train/images/000/01/1104517.jpg</code>. </p>\n<p>The test set images are organized in subfolders <code>test/images/&lt;subfolder&gt;/&lt;image id&gt;.jpg</code>, where <code>&lt;subfolder&gt;</code> corresponds to the integer division of the <code>image_id</code> by 1000. For example, a test image with and <code>image_id</code> of <code>8005</code>, can be found at <code>test/images/008/8005.jpg</code>. </p>",
      "votes": 3,
      "replies": [
        {
          "id": 1265112,
          "author_name": "AndrewTGraham",
          "author_url": "",
          "post_date": "2021-04-06T16:21:23.030000",
          "content": "<p>Thanks a lot! And do the categories correspond to anything of significance, such as species, or are they totally random and just exist to prevent having too many images in one folder which I suppose could be an issue for some computers maybe?</p>\n<p>Additionally, if the categories don't have any significance, are the images in the train folder pre-shuffled?</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 1265768,
          "author_name": "Riccardo de Lutio",
          "author_url": "",
          "post_date": "2021-04-07T07:37:29.623000",
          "content": "<p>Yes categories correspond to species. For the test set however images are shuffled and in different folders such that no folder has too many files. </p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1264494": "Yes, the structure works as follows: \n\nThe training set images are organized in subfolders `train/images/<subfolder1>/<subfolder2>/<image id>.jpg`, where `<subfolder1>` combined with `<subfolder2>` corresponds to the `category_id`. For example, a training image with an `image_id` of `1104517` and a `category_id` of `00001`, can be found at `train/images/000/01/1104517.jpg`. \n\n\nThe test set images are organized in subfolders `test/images/<subfolder>/<image id>.jpg`, where `<subfolder>` corresponds to the integer division of the `image_id` by 1000. For example, a test image with and `image_id` of `8005`, can be found at `test/images/008/8005.jpg`. ",
    "1264146": "Can someone explain the folder system to me. The train images are in numbered folders in numbered folders and the test images are in numbers folders. Why is this? Do the numbers correspond to categories?"
  }
}