{
  "id": 298535,
  "title": "Data-centric: have you noticed unlabeled objects in training set?",
  "url": "/competitions/tensorflow-great-barrier-reef/discussion/298535",
  "author_name": "ademdum",
  "post_date": "2022-01-03T11:59:59.616000",
  "votes": 5,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi community! Watching the amazing notebook <a href=\"https://www.kaggle.com/bamps53/create-annotated-video/\" target=\"_blank\">https://www.kaggle.com/bamps53/create-annotated-video/</a>, and the associated video. I noticed many single COTS are not labeled since the first time captured by the camera, but on some future frames. I think this may confuse our models and increases # of false positives incorrectly.</p>\n<p>If you want check yourself, please see: 06:18 - 06:25 in the previous video for more info.</p>",
  "messages": [
    {
      "id": 1636945,
      "postDate": "2022-01-03T11:59:59.617Z",
      "content": "<p>Hi community! Watching the amazing notebook <a href=\"https://www.kaggle.com/bamps53/create-annotated-video/\" target=\"_blank\">https://www.kaggle.com/bamps53/create-annotated-video/</a>, and the associated video. I noticed many single COTS are not labeled since the first time captured by the camera, but on some future frames. I think this may confuse our models and increases # of false positives incorrectly.</p>\n<p>If you want check yourself, please see: 06:18 - 06:25 in the previous video for more info.</p>",
      "rawMarkdown": "Hi community! Watching the amazing notebook https://www.kaggle.com/bamps53/create-annotated-video/, and the associated video. I noticed many single COTS are not labeled since the first time captured by the camera, but on some future frames. I think this may confuse our models and increases # of false positives incorrectly.\n\nIf you want check yourself, please see: 06:18 - 06:25 in the previous video for more info.",
      "votes": 5
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1636945": "Hi community! Watching the amazing notebook https://www.kaggle.com/bamps53/create-annotated-video/, and the associated video. I noticed many single COTS are not labeled since the first time captured by the camera, but on some future frames. I think this may confuse our models and increases # of false positives incorrectly.\n\nIf you want check yourself, please see: 06:18 - 06:25 in the previous video for more info."
  }
}