{
  "id": 54718,
  "title": "ground truth for stage2 labels",
  "url": "/competitions/data-science-bowl-2018/discussion/54718",
  "author_name": "",
  "post_date": "2018-04-17T04:45:26.493760500Z",
  "votes": 2,
  "comment_count": 3,
  "views": 0,
  "content": "<p>i am still interested to improve my algorithm. I think the images of stage2 can be cluster into few different main cluster types. I wonder if kagglers who are biologists or who those with high scores, or even the organiser can post some ground truth labels for some images in each cluster type.</p>\n\n<p>Because i am not biologist, i have difficulty to know what is the ground truth label, even doing some extensive readup on cell nuclei. For example, how do you label those nerve cells? Some confusion images are shown below:</p>",
  "messages": [
    {
      "id": "315418",
      "postDate": "04/17/2018 04:45:26",
      "content": "<p>i am still interested to improve my algorithm. I think the images of stage2 can be cluster into few different main cluster types. I wonder if kagglers who are biologists or who those with high scores, or even the organiser can post some ground truth labels for some images in each cluster type.</p>\n\n<p>Because i am not biologist, i have difficulty to know what is the ground truth label, even doing some extensive readup on cell nuclei. For example, how do you label those nerve cells? Some confusion images are shown below:</p>",
      "rawMarkdown": "i am still interested to improve my algorithm. I think the images of stage2 can be cluster into few different main cluster types. I wonder if kagglers who are biologists or who those with high scores, or even the organiser can post some ground truth labels for some images in each cluster type.\n\nBecause i am not biologist, i have difficulty to know what is the ground truth label, even doing some extensive readup on cell nuclei. For example, how do you label those nerve cells? Some confusion images are shown below:",
      "votes": null
    },
    {
      "id": "315439",
      "postDate": "04/17/2018 05:42:55",
      "content": "<p>I'm not biologist, but I found they come in pairs, like 527e7-ca63b, ce838-d01ec, 30d1b-aa7a7, ... total 62 pairs. by compare the pairs, I can assume the bright ovals at the crosses are nucleus, not all the smaller dot a net will usually catch. and I submited same predict in pair. detail paris enclosed in .csv.</p>",
      "rawMarkdown": "I'm not biologist, but I found they come in pairs, like 527e7-ca63b, ce838-d01ec, 30d1b-aa7a7, ... total 62 pairs. by compare the pairs, I can assume the bright ovals at the crosses are nucleus, not all the smaller dot a net will usually catch. and I submited same predict in pair. detail paris enclosed in .csv.",
      "votes": null
    },
    {
      "id": "315446",
      "postDate": "04/17/2018 06:01:11",
      "content": "<p>Good finding.</p>",
      "rawMarkdown": "Good finding.",
      "votes": null
    },
    {
      "id": "315738",
      "postDate": "04/17/2018 14:33:52",
      "content": "<p>We could certainly include the metadata if we release the subset of stage 2 images that was scored. See discussion here: <a href=\"https://www.kaggle.com/c/data-science-bowl-2018/discussion/54764\">https://www.kaggle.com/c/data-science-bowl-2018/discussion/54764</a></p>",
      "rawMarkdown": "We could certainly include the metadata if we release the subset of stage 2 images that was scored. See discussion here: https://www.kaggle.com/c/data-science-bowl-2018/discussion/54764",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 315439,
      "author_name": "cranecooker",
      "author_url": "",
      "post_date": "04/17/2018 05:42:55",
      "content": "<p>I'm not biologist, but I found they come in pairs, like 527e7-ca63b, ce838-d01ec, 30d1b-aa7a7, ... total 62 pairs. by compare the pairs, I can assume the bright ovals at the crosses are nucleus, not all the smaller dot a net will usually catch. and I submited same predict in pair. detail paris enclosed in .csv.</p>",
      "votes": null,
      "replies": [
        {
          "id": 315446,
          "author_name": "pliptor",
          "author_url": "",
          "post_date": "04/17/2018 06:01:11",
          "content": "<p>Good finding.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 315738,
      "author_name": "drannecarpenter",
      "author_url": "",
      "post_date": "04/17/2018 14:33:52",
      "content": "<p>We could certainly include the metadata if we release the subset of stage 2 images that was scored. See discussion here: <a href=\"https://www.kaggle.com/c/data-science-bowl-2018/discussion/54764\">https://www.kaggle.com/c/data-science-bowl-2018/discussion/54764</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "315418": "i am still interested to improve my algorithm. I think the images of stage2 can be cluster into few different main cluster types. I wonder if kagglers who are biologists or who those with high scores, or even the organiser can post some ground truth labels for some images in each cluster type.\n\nBecause i am not biologist, i have difficulty to know what is the ground truth label, even doing some extensive readup on cell nuclei. For example, how do you label those nerve cells? Some confusion images are shown below:",
    "315439": "I'm not biologist, but I found they come in pairs, like 527e7-ca63b, ce838-d01ec, 30d1b-aa7a7, ... total 62 pairs. by compare the pairs, I can assume the bright ovals at the crosses are nucleus, not all the smaller dot a net will usually catch. and I submited same predict in pair. detail paris enclosed in .csv.",
    "315446": "Good finding.",
    "315738": "We could certainly include the metadata if we release the subset of stage 2 images that was scored. See discussion here: https://www.kaggle.com/c/data-science-bowl-2018/discussion/54764"
  },
  "source": "meta"
}