{
  "id": 229579,
  "title": "Were 14th grade images necessary in training?",
  "url": "/competitions/vinbigdata-chest-xray-abnormalities-detection/discussion/229579",
  "author_name": "Selman",
  "post_date": "2021-03-30T21:59:07.058000",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Should be included 14th class images  that wasn't illness diagnose in the training? According to me these images wasn't necessary in training . Am I wrong?</p>",
  "messages": [
    {
      "id": 1257467,
      "postDate": "2021-03-30T22:02:44.970Z",
      "content": "<p>We found weird results. Our detector-only model on 15 classes exceeded our 14 class + 2 class classifier pipeline. I guess we don't have to wait too long for the solutions of the top teams though.</p>",
      "rawMarkdown": "We found weird results. Our detector-only model on 15 classes exceeded our 14 class + 2 class classifier pipeline. I guess we don't have to wait too long for the solutions of the top teams though.",
      "votes": 1
    },
    {
      "id": 1257466,
      "postDate": "2021-03-30T21:59:07.060Z",
      "content": "<p>Should be included 14th class images  that wasn't illness diagnose in the training? According to me these images wasn't necessary in training . Am I wrong?</p>",
      "rawMarkdown": "Should be included 14th class images  that wasn't illness diagnose in the training? According to me these images wasn't necessary in training . Am I wrong?",
      "votes": 1
    },
    {
      "id": 1257477,
      "postDate": "2021-03-30T22:22:50.510Z",
      "content": "<p>We used them in some classifiers but usually not in the object detection models. </p>",
      "rawMarkdown": "We used them in some classifiers but usually not in the object detection models. ",
      "votes": 2
    }
  ],
  "comments": [
    {
      "id": 1257467,
      "author_name": "Stanley Zheng",
      "author_url": "",
      "post_date": "2021-03-30T22:02:44.970000",
      "content": "<p>We found weird results. Our detector-only model on 15 classes exceeded our 14 class + 2 class classifier pipeline. I guess we don't have to wait too long for the solutions of the top teams though.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 1257477,
      "author_name": "beluga",
      "author_url": "",
      "post_date": "2021-03-30T22:22:50.510000",
      "content": "<p>We used them in some classifiers but usually not in the object detection models. </p>",
      "votes": 2,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1257467": "We found weird results. Our detector-only model on 15 classes exceeded our 14 class + 2 class classifier pipeline. I guess we don't have to wait too long for the solutions of the top teams though.",
    "1257466": "Should be included 14th class images  that wasn't illness diagnose in the training? According to me these images wasn't necessary in training . Am I wrong?",
    "1257477": "We used them in some classifiers but usually not in the object detection models. "
  }
}