{
  "id": 615300,
  "title": "Understanding mask shape",
  "url": "/competitions/recodai-luc-scientific-image-forgery-detection/discussion/615300",
  "author_name": "",
  "post_date": "2025-11-10T05:17:18.512616100Z",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>I'm not sure what the shape of the masks represent. I see that some masks have shape (1, height, width), which is understandable. But  some masks that have shape (2, height, width), (3, height, width), and I cant figure out what this means. This is my first time doing segmentation, and I don't understand what the first dimension denotes. Any help would be appreciated. </p>",
  "messages": [
    {
      "id": "3314156",
      "postDate": "11/10/2025 05:17:18",
      "content": "<p>I'm not sure what the shape of the masks represent. I see that some masks have shape (1, height, width), which is understandable. But  some masks that have shape (2, height, width), (3, height, width), and I cant figure out what this means. This is my first time doing segmentation, and I don't understand what the first dimension denotes. Any help would be appreciated. </p>",
      "rawMarkdown": "I'm not sure what the shape of the masks represent. I see that some masks have shape (1, height, width), which is understandable. But  some masks that have shape (2, height, width), (3, height, width), and I cant figure out what this means. This is my first time doing segmentation, and I don't understand what the first dimension denotes. Any help would be appreciated.",
      "votes": null
    },
    {
      "id": "3314188",
      "postDate": "11/10/2025 05:36:37",
      "content": "<p>it may mean the forgery regions have multi-source, each channel include one forgery region.</p>",
      "rawMarkdown": "it may mean the forgery regions have multi-source, each channel include one forgery region.",
      "votes": null
    },
    {
      "id": "3314208",
      "postDate": "11/10/2025 05:45:37",
      "content": "<p>oooo, that makes so much sense! Thanks a lot!</p>",
      "rawMarkdown": "oooo, that makes so much sense! Thanks a lot!",
      "votes": null
    },
    {
      "id": "3314983",
      "postDate": "11/10/2025 12:55:40",
      "content": "<p>That's exactly what it means. Each channel refers to a different forgery region.</p>\n<p>For instance, if cell A and cell B are both copied:</p>\n<ul>\n<li>Channel 1 will contain the mask for cell A and all of its copies.</li>\n<li>Channel 2 will contain the mask for cell B and all of its copies.</li>\n</ul>",
      "rawMarkdown": "That's exactly what it means. Each channel refers to a different forgery region.\n\nFor instance, if cell A and cell B are both copied:\n- Channel 1 will contain the mask for cell A and all of its copies.\n- Channel 2 will contain the mask for cell B and all of its copies.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3314188,
      "author_name": "bobtony",
      "author_url": "",
      "post_date": "11/10/2025 05:36:37",
      "content": "<p>it may mean the forgery regions have multi-source, each channel include one forgery region.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3314208,
          "author_name": "jumpyjamie",
          "author_url": "",
          "post_date": "11/10/2025 05:45:37",
          "content": "<p>oooo, that makes so much sense! Thanks a lot!</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 3314983,
          "author_name": "joophillipecardenuto",
          "author_url": "",
          "post_date": "11/10/2025 12:55:40",
          "content": "<p>That's exactly what it means. Each channel refers to a different forgery region.</p>\n<p>For instance, if cell A and cell B are both copied:</p>\n<ul>\n<li>Channel 1 will contain the mask for cell A and all of its copies.</li>\n<li>Channel 2 will contain the mask for cell B and all of its copies.</li>\n</ul>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3314156": "I'm not sure what the shape of the masks represent. I see that some masks have shape (1, height, width), which is understandable. But  some masks that have shape (2, height, width), (3, height, width), and I cant figure out what this means. This is my first time doing segmentation, and I don't understand what the first dimension denotes. Any help would be appreciated.",
    "3314188": "it may mean the forgery regions have multi-source, each channel include one forgery region.",
    "3314208": "oooo, that makes so much sense! Thanks a lot!",
    "3314983": "That's exactly what it means. Each channel refers to a different forgery region.\n\nFor instance, if cell A and cell B are both copied:\n- Channel 1 will contain the mask for cell A and all of its copies.\n- Channel 2 will contain the mask for cell B and all of its copies."
  },
  "source": "meta"
}