{
  "id": 230669,
  "title": "Instance Segmentation from Semantic Segmentation?",
  "url": "/competitions/hpa-single-cell-image-classification/discussion/230669",
  "author_name": "Bo Peng",
  "post_date": "2021-04-05T06:37:03.646000",
  "votes": 3,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi,</p>\n<p>This is not specific to this competition but do you guys know of any good techniques from getting instance segmentation masks from a dataset that is labeled using semantic segmentation?</p>\n<p>So if this was a semantic segmentation dataset, how would we take out the individual instances of nuclei/cells if needed?</p>\n<p>Thanks for the help,</p>",
  "messages": [
    {
      "id": 1263175,
      "postDate": "2021-04-05T06:37:03.647Z",
      "content": "<p>Hi,</p>\n<p>This is not specific to this competition but do you guys know of any good techniques from getting instance segmentation masks from a dataset that is labeled using semantic segmentation?</p>\n<p>So if this was a semantic segmentation dataset, how would we take out the individual instances of nuclei/cells if needed?</p>\n<p>Thanks for the help,</p>",
      "rawMarkdown": "Hi,\n\nThis is not specific to this competition but do you guys know of any good techniques from getting instance segmentation masks from a dataset that is labeled using semantic segmentation?\n\nSo if this was a semantic segmentation dataset, how would we take out the individual instances of nuclei/cells if needed?\n\nThanks for the help,",
      "votes": 3
    },
    {
      "id": 1263435,
      "postDate": "2021-04-05T12:16:22.200Z",
      "content": "<p>Hi, one way is using morphological postprocessing on semantic masks. This is already done in hpacellsegmentator.</p>",
      "rawMarkdown": "Hi, one way is using morphological postprocessing on semantic masks. This is already done in hpacellsegmentator.",
      "votes": 1,
      "replies": [
        {
          "id": 1263782,
          "postDate": "2021-04-05T17:10:20.090Z",
          "rawMarkdown": "",
          "isDeleted": true
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 1263435,
      "author_name": "Vladislav Ostankovich",
      "author_url": "",
      "post_date": "2021-04-05T12:16:22.200000",
      "content": "<p>Hi, one way is using morphological postprocessing on semantic masks. This is already done in hpacellsegmentator.</p>",
      "votes": 1,
      "replies": [
        {
          "id": 1263782,
          "author_name": "",
          "author_url": "",
          "post_date": "2021-04-05T17:10:20.090000",
          "content": "",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1263175": "Hi,\n\nThis is not specific to this competition but do you guys know of any good techniques from getting instance segmentation masks from a dataset that is labeled using semantic segmentation?\n\nSo if this was a semantic segmentation dataset, how would we take out the individual instances of nuclei/cells if needed?\n\nThanks for the help,",
    "1263435": "Hi, one way is using morphological postprocessing on semantic masks. This is already done in hpacellsegmentator."
  }
}