{
  "id": 419190,
  "title": "How to generate proper bounding boxes from mask when objects are merged?",
  "url": "/competitions/google-research-identify-contrails-reduce-global-warming/discussion/419190",
  "author_name": "Giri Prasad512",
  "post_date": "2023-06-24T16:42:29.411000",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F10628427%2F162d0d012b0913539ec8824e4b833dfa%2FScreenshot%20from%202023-06-24%2021-58-45.png?generation=1687624893075288&amp;alt=media\" alt=\"\"></p>",
  "messages": [
    {
      "id": 2316383,
      "postDate": "2023-06-24T21:00:50.517Z",
      "content": "<p>I don't think there is a way to train on this dataset with bounding boxes since objects completely overlap with contrails within the bounding box of other contrails(please correct me if I'm wrong). Especially when you start considering nms (non max suppression) which is used in models with bounding boxes since nms removes bounding boxes based on a certain level of overlap and removes the less confident ones.</p>",
      "rawMarkdown": "I don't think there is a way to train on this dataset with bounding boxes since objects completely overlap with contrails within the bounding box of other contrails(please correct me if I'm wrong). Especially when you start considering nms (non max suppression) which is used in models with bounding boxes since nms removes bounding boxes based on a certain level of overlap and removes the less confident ones.",
      "votes": 2
    },
    {
      "id": 2316108,
      "postDate": "2023-06-24T16:42:29.410Z",
      "content": "<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F10628427%2F162d0d012b0913539ec8824e4b833dfa%2FScreenshot%20from%202023-06-24%2021-58-45.png?generation=1687624893075288&amp;alt=media\" alt=\"\"></p>",
      "rawMarkdown": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F10628427%2F162d0d012b0913539ec8824e4b833dfa%2FScreenshot%20from%202023-06-24%2021-58-45.png?generation=1687624893075288&alt=media)",
      "votes": 2
    }
  ],
  "comments": [
    {
      "id": 2316383,
      "author_name": "Ari",
      "author_url": "",
      "post_date": "2023-06-24T21:00:50.517000",
      "content": "<p>I don't think there is a way to train on this dataset with bounding boxes since objects completely overlap with contrails within the bounding box of other contrails(please correct me if I'm wrong). Especially when you start considering nms (non max suppression) which is used in models with bounding boxes since nms removes bounding boxes based on a certain level of overlap and removes the less confident ones.</p>",
      "votes": 2,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2316383": "I don't think there is a way to train on this dataset with bounding boxes since objects completely overlap with contrails within the bounding box of other contrails(please correct me if I'm wrong). Especially when you start considering nms (non max suppression) which is used in models with bounding boxes since nms removes bounding boxes based on a certain level of overlap and removes the less confident ones.",
    "2316108": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F10628427%2F162d0d012b0913539ec8824e4b833dfa%2FScreenshot%20from%202023-06-24%2021-58-45.png?generation=1687624893075288&alt=media)"
  }
}