{
  "id": 532982,
  "title": "Understanding Image Segmentation with Limited Coordinate Data",
  "url": "/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/532982",
  "author_name": "",
  "post_date": "2024-09-09T03:24:20.143983900Z",
  "votes": 1,
  "comment_count": 7,
  "views": 0,
  "content": "<p>Hi everyone,</p>\n<p>I’ve come across this challenging competition. I found that the labels provided only include one X coordinate and one Y coordinate per object. I’m struggling to understand how segmentation can be effectively performed with such limited data.</p>\n<p>Does anyone can give some clues on how to utilize a single coordinate for segmentation?<br>\nLooking forward to your insights and experiences!</p>\n<p>Best <br>\nLeo</p>",
  "messages": [
    {
      "id": "2983686",
      "postDate": "09/09/2024 03:24:20",
      "content": "<p>Hi everyone,</p>\n<p>I’ve come across this challenging competition. I found that the labels provided only include one X coordinate and one Y coordinate per object. I’m struggling to understand how segmentation can be effectively performed with such limited data.</p>\n<p>Does anyone can give some clues on how to utilize a single coordinate for segmentation?<br>\nLooking forward to your insights and experiences!</p>\n<p>Best <br>\nLeo</p>",
      "rawMarkdown": "Hi everyone,\n\nI’ve come across this challenging competition. I found that the labels provided only include one X coordinate and one Y coordinate per object. I’m struggling to understand how segmentation can be effectively performed with such limited data.\n\nDoes anyone can give some clues on how to utilize a single coordinate for segmentation?\nLooking forward to your insights and experiences!\n\nBest \nLeo",
      "votes": null
    },
    {
      "id": "2983847",
      "postDate": "09/09/2024 06:52:46",
      "content": "<p>I am a beginner in image processing competitions. From the discussions, I understand that segmentation is an effective approach. In this competition, how exactly is segmentation being utilized? Is it used to crop only the relevant area or change its color, for example? I may be asking something off the mark since I'm not very familiar with this, but I would appreciate it if you could answer.</p>",
      "rawMarkdown": "I am a beginner in image processing competitions. From the discussions, I understand that segmentation is an effective approach. In this competition, how exactly is segmentation being utilized? Is it used to crop only the relevant area or change its color, for example? I may be asking something off the mark since I'm not very familiar with this, but I would appreciate it if you could answer.",
      "votes": null
    },
    {
      "id": "2984218",
      "postDate": "09/09/2024 13:28:40",
      "content": "<p>I had a look at the 2023 winners solution. I thinks this is generally a 2-stage task. </p>\n<ol>\n<li>3d segmentation to find out the related part (how the target is defined since there is only one x and y coordinate ?). </li>\n<li>Image classification.</li>\n</ol>",
      "rawMarkdown": "I had a look at the 2023 winners solution. I thinks this is generally a 2-stage task. \n1. 3d segmentation to find out the related part (how the target is defined since there is only one x and y coordinate ?). \n2. Image classification.",
      "votes": null
    },
    {
      "id": "2984583",
      "postDate": "09/09/2024 18:25:24",
      "content": "<p>there is a public notebook demoing sam2. sam2 is capable of segmenting anything with a single coordinate.</p>",
      "rawMarkdown": "there is a public notebook demoing sam2. sam2 is capable of segmenting anything with a single coordinate.",
      "votes": null
    },
    {
      "id": "2984806",
      "postDate": "09/10/2024 02:36:15",
      "content": "<p>Oh really. Thanks for the info! Is it on Kaggle</p>",
      "rawMarkdown": "Oh really. Thanks for the info! Is it on Kaggle",
      "votes": null
    },
    {
      "id": "2984812",
      "postDate": "09/10/2024 02:50:09",
      "content": "<p><a href=\"https://www.kaggle.com/code/claverru/automatic-spine-cord-segmentation-sam-2\" target=\"_blank\">https://www.kaggle.com/code/claverru/automatic-spine-cord-segmentation-sam-2</a></p>\n<p>the demo uses 5 points. i tried with one point and it still does the same segmentation.</p>",
      "rawMarkdown": "https://www.kaggle.com/code/claverru/automatic-spine-cord-segmentation-sam-2\n\nthe demo uses 5 points. i tried with one point and it still does the same segmentation.",
      "votes": null
    },
    {
      "id": "2984815",
      "postDate": "09/10/2024 02:55:38",
      "content": "<p>Did this segmentation improve accuracy?</p>",
      "rawMarkdown": "Did this segmentation improve accuracy?",
      "votes": null
    },
    {
      "id": "2984825",
      "postDate": "09/10/2024 03:06:04",
      "content": "<p>Exactly. I saw it too.<br>\nI think we need to seg 5 points to 5 masks</p>",
      "rawMarkdown": "Exactly. I saw it too.\nI think we need to seg 5 points to 5 masks",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2983847,
      "author_name": "mukatai",
      "author_url": "",
      "post_date": "09/09/2024 06:52:46",
      "content": "<p>I am a beginner in image processing competitions. From the discussions, I understand that segmentation is an effective approach. In this competition, how exactly is segmentation being utilized? Is it used to crop only the relevant area or change its color, for example? I may be asking something off the mark since I'm not very familiar with this, but I would appreciate it if you could answer.</p>",
      "votes": null,
      "replies": [
        {
          "id": 2984218,
          "author_name": "yxyyxy",
          "author_url": "",
          "post_date": "09/09/2024 13:28:40",
          "content": "<p>I had a look at the 2023 winners solution. I thinks this is generally a 2-stage task. </p>\n<ol>\n<li>3d segmentation to find out the related part (how the target is defined since there is only one x and y coordinate ?). </li>\n<li>Image classification.</li>\n</ol>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 2984583,
      "author_name": "ibinti",
      "author_url": "",
      "post_date": "09/09/2024 18:25:24",
      "content": "<p>there is a public notebook demoing sam2. sam2 is capable of segmenting anything with a single coordinate.</p>",
      "votes": null,
      "replies": [
        {
          "id": 2984806,
          "author_name": "yxyyxy",
          "author_url": "",
          "post_date": "09/10/2024 02:36:15",
          "content": "<p>Oh really. Thanks for the info! Is it on Kaggle</p>",
          "votes": null,
          "replies": [
            {
              "id": 2984812,
              "author_name": "ibinti",
              "author_url": "",
              "post_date": "09/10/2024 02:50:09",
              "content": "<p><a href=\"https://www.kaggle.com/code/claverru/automatic-spine-cord-segmentation-sam-2\" target=\"_blank\">https://www.kaggle.com/code/claverru/automatic-spine-cord-segmentation-sam-2</a></p>\n<p>the demo uses 5 points. i tried with one point and it still does the same segmentation.</p>",
              "votes": null,
              "replies": [
                {
                  "id": 2984815,
                  "author_name": "mukatai",
                  "author_url": "",
                  "post_date": "09/10/2024 02:55:38",
                  "content": "<p>Did this segmentation improve accuracy?</p>",
                  "votes": null,
                  "replies": []
                },
                {
                  "id": 2984825,
                  "author_name": "yxyyxy",
                  "author_url": "",
                  "post_date": "09/10/2024 03:06:04",
                  "content": "<p>Exactly. I saw it too.<br>\nI think we need to seg 5 points to 5 masks</p>",
                  "votes": null,
                  "replies": []
                }
              ]
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2983686": "Hi everyone,\n\nI’ve come across this challenging competition. I found that the labels provided only include one X coordinate and one Y coordinate per object. I’m struggling to understand how segmentation can be effectively performed with such limited data.\n\nDoes anyone can give some clues on how to utilize a single coordinate for segmentation?\nLooking forward to your insights and experiences!\n\nBest \nLeo",
    "2983847": "I am a beginner in image processing competitions. From the discussions, I understand that segmentation is an effective approach. In this competition, how exactly is segmentation being utilized? Is it used to crop only the relevant area or change its color, for example? I may be asking something off the mark since I'm not very familiar with this, but I would appreciate it if you could answer.",
    "2984218": "I had a look at the 2023 winners solution. I thinks this is generally a 2-stage task. \n1. 3d segmentation to find out the related part (how the target is defined since there is only one x and y coordinate ?). \n2. Image classification.",
    "2984583": "there is a public notebook demoing sam2. sam2 is capable of segmenting anything with a single coordinate.",
    "2984806": "Oh really. Thanks for the info! Is it on Kaggle",
    "2984812": "https://www.kaggle.com/code/claverru/automatic-spine-cord-segmentation-sam-2\n\nthe demo uses 5 points. i tried with one point and it still does the same segmentation.",
    "2984815": "Did this segmentation improve accuracy?",
    "2984825": "Exactly. I saw it too.\nI think we need to seg 5 points to 5 masks"
  },
  "source": "meta"
}