{
  "id": 154774,
  "title": "Tiny bboxes",
  "url": "/competitions/global-wheat-detection/discussion/154774",
  "author_name": "",
  "post_date": "2020-05-29T19:13:42.524581Z",
  "votes": 2,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Hi everyone!</p>\n\n<p>I found some tiny bboxes in images as following:\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1965756%2Fff5213ee07c6f3f01749009292ac1c9d%2F2020-05-30%204.04.15.png?generation=1590779230187379&amp;alt=media\" alt=\"\"></p>\n\n<p>Which method should I use for training?\n    a. Remove from training data.\n    b. Label as 'background'.\n    c. Other.</p>\n\n<p>Thank you.</p>",
  "messages": [
    {
      "id": "866902",
      "postDate": "05/29/2020 19:13:42",
      "content": "<p>Hi everyone!</p>\n\n<p>I found some tiny bboxes in images as following:\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1965756%2Fff5213ee07c6f3f01749009292ac1c9d%2F2020-05-30%204.04.15.png?generation=1590779230187379&amp;alt=media\" alt=\"\"></p>\n\n<p>Which method should I use for training?\n    a. Remove from training data.\n    b. Label as 'background'.\n    c. Other.</p>\n\n<p>Thank you.</p>",
      "rawMarkdown": "Hi everyone!\n\nI found some tiny bboxes in images as following:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1965756%2Fff5213ee07c6f3f01749009292ac1c9d%2F2020-05-30%204.04.15.png?generation=1590779230187379&amp;alt=media)\n\nWhich method should I use for training?\n    a. Remove from training data.\n    b. Label as 'background'.\n    c. Other.\n\nThank you.",
      "votes": null
    },
    {
      "id": "867062",
      "postDate": "05/30/2020 01:00:44",
      "content": "<p>Just remove it from the data. In general I think I filter boxes smaller than 400 square pixels in area</p>",
      "rawMarkdown": "Just remove it from the data. In general I think I filter boxes smaller than 400 square pixels in area",
      "votes": null
    },
    {
      "id": "868041",
      "postDate": "05/30/2020 20:38:32",
      "content": "<p>From the competition host:</p>\n\n<blockquote>\n  <p>Another issue I've noticed working on that there are some too small boxes are present in the file. I thought I filtered them before but apparently not. This small boxes often have a height or width below 10 and was too difficult to notice by a human reviewer on our tool. Most of the problems will be solved in the update and again it represents very few bounding boxes (apparently 170 on almost 150k). </p>\n</blockquote>",
      "rawMarkdown": "From the competition host:\n&gt; Another issue I've noticed working on that there are some too small boxes are present in the file. I thought I filtered them before but apparently not. This small boxes often have a height or width below 10 and was too difficult to notice by a human reviewer on our tool. Most of the problems will be solved in the update and again it represents very few bounding boxes (apparently 170 on almost 150k).",
      "votes": null
    },
    {
      "id": "868434",
      "postDate": "05/31/2020 08:25:13",
      "content": "<p>Thank you for your information!</p>",
      "rawMarkdown": "Thank you for your information!",
      "votes": null
    },
    {
      "id": "868436",
      "postDate": "05/31/2020 08:25:49",
      "content": "<p>Thank you!\nI'll try it.</p>",
      "rawMarkdown": "Thank you!\nI'll try it.",
      "votes": null
    },
    {
      "id": "868597",
      "postDate": "05/31/2020 11:11:25",
      "content": "<p>I also remove it. I filtered boxes smaller than 10 pixels in width and height</p>",
      "rawMarkdown": "I also remove it. I filtered boxes smaller than 10 pixels in width and height",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 867062,
      "author_name": "matthewmasters",
      "author_url": "",
      "post_date": "05/30/2020 01:00:44",
      "content": "<p>Just remove it from the data. In general I think I filter boxes smaller than 400 square pixels in area</p>",
      "votes": null,
      "replies": [
        {
          "id": 868436,
          "author_name": "sakumatcho",
          "author_url": "",
          "post_date": "05/31/2020 08:25:49",
          "content": "<p>Thank you!\nI'll try it.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 868041,
      "author_name": "sergeyzlobin",
      "author_url": "",
      "post_date": "05/30/2020 20:38:32",
      "content": "<p>From the competition host:</p>\n\n<blockquote>\n  <p>Another issue I've noticed working on that there are some too small boxes are present in the file. I thought I filtered them before but apparently not. This small boxes often have a height or width below 10 and was too difficult to notice by a human reviewer on our tool. Most of the problems will be solved in the update and again it represents very few bounding boxes (apparently 170 on almost 150k). </p>\n</blockquote>",
      "votes": null,
      "replies": [
        {
          "id": 868434,
          "author_name": "sakumatcho",
          "author_url": "",
          "post_date": "05/31/2020 08:25:13",
          "content": "<p>Thank you for your information!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 868597,
      "author_name": "comojin1994",
      "author_url": "",
      "post_date": "05/31/2020 11:11:25",
      "content": "<p>I also remove it. I filtered boxes smaller than 10 pixels in width and height</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "866902": "Hi everyone!\n\nI found some tiny bboxes in images as following:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1965756%2Fff5213ee07c6f3f01749009292ac1c9d%2F2020-05-30%204.04.15.png?generation=1590779230187379&amp;alt=media)\n\nWhich method should I use for training?\n    a. Remove from training data.\n    b. Label as 'background'.\n    c. Other.\n\nThank you.",
    "867062": "Just remove it from the data. In general I think I filter boxes smaller than 400 square pixels in area",
    "868041": "From the competition host:\n&gt; Another issue I've noticed working on that there are some too small boxes are present in the file. I thought I filtered them before but apparently not. This small boxes often have a height or width below 10 and was too difficult to notice by a human reviewer on our tool. Most of the problems will be solved in the update and again it represents very few bounding boxes (apparently 170 on almost 150k).",
    "868434": "Thank you for your information!",
    "868436": "Thank you!\nI'll try it.",
    "868597": "I also remove it. I filtered boxes smaller than 10 pixels in width and height"
  },
  "source": "meta"
}