{
  "id": 464104,
  "title": "For anyone who is getting a score of 0/scoring error",
  "url": "/competitions/blood-vessel-segmentation/discussion/464104",
  "author_name": "zUni8789798",
  "post_date": "2023-12-28T19:03:00.163000",
  "votes": 2,
  "comment_count": 0,
  "views": 0,
  "content": "<p>For the scoring of 0, make sure that you binarize your submissions. For some reason, if you don't your submissions will be full of false positives and your empty masks will become bogus RLE encodes. </p>\n<p>For invalid submissions, I highly recommend you plot the masks your model is predicting. If your masks are wild and all over the place, this will cause a submission error. Make sure your submission masks for the sample test set are empty or very minimal. Hope this helps some of you.</p>",
  "messages": [
    {
      "id": 2577744,
      "postDate": "2023-12-28T19:03:00.163Z",
      "content": "<p>For the scoring of 0, make sure that you binarize your submissions. For some reason, if you don't your submissions will be full of false positives and your empty masks will become bogus RLE encodes. </p>\n<p>For invalid submissions, I highly recommend you plot the masks your model is predicting. If your masks are wild and all over the place, this will cause a submission error. Make sure your submission masks for the sample test set are empty or very minimal. Hope this helps some of you.</p>",
      "rawMarkdown": "For the scoring of 0, make sure that you binarize your submissions. For some reason, if you don't your submissions will be full of false positives and your empty masks will become bogus RLE encodes. \n\nFor invalid submissions, I highly recommend you plot the masks your model is predicting. If your masks are wild and all over the place, this will cause a submission error. Make sure your submission masks for the sample test set are empty or very minimal. Hope this helps some of you.\n\n",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2577744": "For the scoring of 0, make sure that you binarize your submissions. For some reason, if you don't your submissions will be full of false positives and your empty masks will become bogus RLE encodes. \n\nFor invalid submissions, I highly recommend you plot the masks your model is predicting. If your masks are wild and all over the place, this will cause a submission error. Make sure your submission masks for the sample test set are empty or very minimal. Hope this helps some of you.\n\n"
  }
}