{
  "id": 247909,
  "title": "HI,I have some doubt about this competition?",
  "url": "/competitions/siim-covid19-detection/discussion/247909",
  "author_name": "",
  "post_date": "2021-06-21T17:14:42.008190400Z",
  "votes": 2,
  "comment_count": 3,
  "views": 0,
  "content": "<p>we need to trained two models to predict result? The one is the objection detection model to detect the bbox and classification. The other one model is to classification the study level result? So ,we need to have two models? Thanks!</p>",
  "messages": [
    {
      "id": "1359952",
      "postDate": "06/21/2021 17:14:42",
      "content": "<p>we need to trained two models to predict result? The one is the objection detection model to detect the bbox and classification. The other one model is to classification the study level result? So ,we need to have two models? Thanks!</p>",
      "rawMarkdown": "we need to trained two models to predict result? The one is the objection detection model to detect the bbox and classification. The other one model is to classification the study level result? So ,we need to have two models? Thanks!",
      "votes": null
    },
    {
      "id": "1394837",
      "postDate": "07/20/2021 16:59:20",
      "content": "<p>Hi I am new here as well. Have you figured out whether we need two models or not? I am curious about this as well</p>",
      "rawMarkdown": "Hi I am new here as well. Have you figured out whether we need two models or not? I am curious about this as well",
      "votes": null
    },
    {
      "id": "1394868",
      "postDate": "07/20/2021 17:30:50",
      "content": "<p>Hey, it depends on what approach you're using. You may use either 2 models or 3 models based on your necessity. Usually, most of the public notebooks you could see use 3 models for submission which include binary classifier, multiclass classifier, and object detection. You could also try some alternative ways to remove the need for a binary classifier for predicting confidence on whether an image has BBox or not. </p>",
      "rawMarkdown": "Hey, it depends on what approach you're using. You may use either 2 models or 3 models based on your necessity. Usually, most of the public notebooks you could see use 3 models for submission which include binary classifier, multiclass classifier, and object detection. You could also try some alternative ways to remove the need for a binary classifier for predicting confidence on whether an image has BBox or not.",
      "votes": null
    },
    {
      "id": "1395065",
      "postDate": "07/20/2021 21:24:48",
      "content": "<p>Thank you very much for this! I also noticed that the one of the code requirements is to have run time below 9 hours. If that is the case, there will not be much time for each model to be trained right?</p>",
      "rawMarkdown": "Thank you very much for this! I also noticed that the one of the code requirements is to have run time below 9 hours. If that is the case, there will not be much time for each model to be trained right?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1394837,
      "author_name": "feanor007",
      "author_url": "",
      "post_date": "07/20/2021 16:59:20",
      "content": "<p>Hi I am new here as well. Have you figured out whether we need two models or not? I am curious about this as well</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1394868,
      "author_name": "nischaydnk",
      "author_url": "",
      "post_date": "07/20/2021 17:30:50",
      "content": "<p>Hey, it depends on what approach you're using. You may use either 2 models or 3 models based on your necessity. Usually, most of the public notebooks you could see use 3 models for submission which include binary classifier, multiclass classifier, and object detection. You could also try some alternative ways to remove the need for a binary classifier for predicting confidence on whether an image has BBox or not. </p>",
      "votes": null,
      "replies": [
        {
          "id": 1395065,
          "author_name": "feanor007",
          "author_url": "",
          "post_date": "07/20/2021 21:24:48",
          "content": "<p>Thank you very much for this! I also noticed that the one of the code requirements is to have run time below 9 hours. If that is the case, there will not be much time for each model to be trained right?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1359952": "we need to trained two models to predict result? The one is the objection detection model to detect the bbox and classification. The other one model is to classification the study level result? So ,we need to have two models? Thanks!",
    "1394837": "Hi I am new here as well. Have you figured out whether we need two models or not? I am curious about this as well",
    "1394868": "Hey, it depends on what approach you're using. You may use either 2 models or 3 models based on your necessity. Usually, most of the public notebooks you could see use 3 models for submission which include binary classifier, multiclass classifier, and object detection. You could also try some alternative ways to remove the need for a binary classifier for predicting confidence on whether an image has BBox or not.",
    "1395065": "Thank you very much for this! I also noticed that the one of the code requirements is to have run time below 9 hours. If that is the case, there will not be much time for each model to be trained right?"
  },
  "source": "meta"
}