{
  "id": 295689,
  "title": "Can anyone tell me the situation you met when \"Submission Scoring Error\" occurred?",
  "url": "/competitions/sartorius-cell-instance-segmentation/discussion/295689",
  "author_name": "",
  "post_date": "2021-12-17T09:31:34.103107Z",
  "votes": 3,
  "comment_count": 6,
  "views": 0,
  "content": "<p>I have been stuck in “Submission Scoring Error” for many days.  I have checked my code again and again, but all failed. Limited by GPU source, I just split the whole image into four overlapping patches, then merged results with NMS during postprocessing. When testing, I did the same process but always got \"Submission Scoring Error\". Howerer, when I fed the whole image into model during testing, it worked well. That really made me confused.😭</p>\n<p>So can anyone tell me the situation you met when \"Submission Scoring Error\" occurred? Why it returns this error? Please give me some ideas to solve it! Please!😭</p>\n<p>Many Thanks!</p>",
  "messages": [
    {
      "id": "1620956",
      "postDate": "12/17/2021 09:31:34",
      "content": "<p>I have been stuck in “Submission Scoring Error” for many days.  I have checked my code again and again, but all failed. Limited by GPU source, I just split the whole image into four overlapping patches, then merged results with NMS during postprocessing. When testing, I did the same process but always got \"Submission Scoring Error\". Howerer, when I fed the whole image into model during testing, it worked well. That really made me confused.😭</p>\n<p>So can anyone tell me the situation you met when \"Submission Scoring Error\" occurred? Why it returns this error? Please give me some ideas to solve it! Please!😭</p>\n<p>Many Thanks!</p>",
      "rawMarkdown": "I have been stuck in “Submission Scoring Error” for many days.  I have checked my code again and again, but all failed. Limited by GPU source, I just split the whole image into four overlapping patches, then merged results with NMS during postprocessing. When testing, I did the same process but always got \"Submission Scoring Error\". Howerer, when I fed the whole image into model during testing, it worked well. That really made me confused.😭\n\nSo can anyone tell me the situation you met when \"Submission Scoring Error\" occurred? Why it returns this error? Please give me some ideas to solve it! Please!😭\n\nMany Thanks!",
      "votes": null
    },
    {
      "id": "1621261",
      "postDate": "12/17/2021 15:19:24",
      "content": "<p>One possible cause is mask not being detected at all.</p>",
      "rawMarkdown": "One possible cause is mask not being detected at all.",
      "votes": null
    },
    {
      "id": "1621496",
      "postDate": "12/17/2021 18:44:18",
      "content": "<p>I have the same problem and I have been working on it for 3 days (not counting the days to understand how to load the damned packages offline):</p>\n<p>a. no mask at all for a particular id<br>\nb. mask somehow rotated 90 deg or resized h,w instead of w,h…<br>\nc. …</p>",
      "rawMarkdown": "I have the same problem and I have been working on it for 3 days (not counting the days to understand how to load the damned packages offline):\n\na. no mask at all for a particular id\nb. mask somehow rotated 90 deg or resized h,w instead of w,h...\nc. ...",
      "votes": null
    },
    {
      "id": "1622186",
      "postDate": "12/18/2021 12:58:01",
      "content": "<p>Thanks for your ideas! I will try to construct mask manually to verify problem a. What a damned problem!😓</p>",
      "rawMarkdown": "Thanks for your ideas! I will try to construct mask manually to verify problem a. What a damned problem!😓",
      "votes": null
    },
    {
      "id": "1622189",
      "postDate": "12/18/2021 13:00:52",
      "content": "<p>Got it. I haven't considered this issue before. Thanks a lot!</p>",
      "rawMarkdown": "Got it. I haven't considered this issue before. Thanks a lot!",
      "votes": null
    },
    {
      "id": "1623523",
      "postDate": "12/20/2021 01:48:04",
      "content": "<p>I have set the bbox confidence threshold to zero, in this case, the notebook worked well. So the problem said above that \"no mask at all for a particular id\" was true.</p>",
      "rawMarkdown": "I have set the bbox confidence threshold to zero, in this case, the notebook worked well. So the problem said above that \"no mask at all for a particular id\" was true.",
      "votes": null
    },
    {
      "id": "1625437",
      "postDate": "12/21/2021 19:37:05",
      "content": "<p>If the masks you're predicting overlap that wuld throw an error</p>",
      "rawMarkdown": "If the masks you're predicting overlap that wuld throw an error",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1621261,
      "author_name": "atamazian",
      "author_url": "",
      "post_date": "12/17/2021 15:19:24",
      "content": "<p>One possible cause is mask not being detected at all.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1622189,
          "author_name": "wangyuanxiang",
          "author_url": "",
          "post_date": "12/18/2021 13:00:52",
          "content": "<p>Got it. I haven't considered this issue before. Thanks a lot!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1621496,
      "author_name": "chabir",
      "author_url": "",
      "post_date": "12/17/2021 18:44:18",
      "content": "<p>I have the same problem and I have been working on it for 3 days (not counting the days to understand how to load the damned packages offline):</p>\n<p>a. no mask at all for a particular id<br>\nb. mask somehow rotated 90 deg or resized h,w instead of w,h…<br>\nc. …</p>",
      "votes": null,
      "replies": [
        {
          "id": 1622186,
          "author_name": "wangyuanxiang",
          "author_url": "",
          "post_date": "12/18/2021 12:58:01",
          "content": "<p>Thanks for your ideas! I will try to construct mask manually to verify problem a. What a damned problem!😓</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1623523,
      "author_name": "wangyuanxiang",
      "author_url": "",
      "post_date": "12/20/2021 01:48:04",
      "content": "<p>I have set the bbox confidence threshold to zero, in this case, the notebook worked well. So the problem said above that \"no mask at all for a particular id\" was true.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1625437,
      "author_name": "ksmcg90",
      "author_url": "",
      "post_date": "12/21/2021 19:37:05",
      "content": "<p>If the masks you're predicting overlap that wuld throw an error</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1620956": "I have been stuck in “Submission Scoring Error” for many days.  I have checked my code again and again, but all failed. Limited by GPU source, I just split the whole image into four overlapping patches, then merged results with NMS during postprocessing. When testing, I did the same process but always got \"Submission Scoring Error\". Howerer, when I fed the whole image into model during testing, it worked well. That really made me confused.😭\n\nSo can anyone tell me the situation you met when \"Submission Scoring Error\" occurred? Why it returns this error? Please give me some ideas to solve it! Please!😭\n\nMany Thanks!",
    "1621261": "One possible cause is mask not being detected at all.",
    "1621496": "I have the same problem and I have been working on it for 3 days (not counting the days to understand how to load the damned packages offline):\n\na. no mask at all for a particular id\nb. mask somehow rotated 90 deg or resized h,w instead of w,h...\nc. ...",
    "1622186": "Thanks for your ideas! I will try to construct mask manually to verify problem a. What a damned problem!😓",
    "1622189": "Got it. I haven't considered this issue before. Thanks a lot!",
    "1623523": "I have set the bbox confidence threshold to zero, in this case, the notebook worked well. So the problem said above that \"no mask at all for a particular id\" was true.",
    "1625437": "If the masks you're predicting overlap that wuld throw an error"
  },
  "source": "meta"
}