{
  "id": 664426,
  "title": "Mask data RGB versus 3 separate cells?",
  "url": "/competitions/recodai-luc-scientific-image-forgery-detection/discussion/664426",
  "author_name": "",
  "post_date": "2025-12-25T01:22:52.412803400Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Any one know how to identify the below </p>\n<ol>\n<li>3D mask (3,h,w) layers when there are 3 different forged cells (one for each layer,) vs RGB data?\nThe numpy array header data is not useful as far as I know:\ne.g: {'descr': '|u1', 'fortran_order': False, 'shape': (3, 649, 884), }</li>\n<li>2D Numpy arrays (2,h,w) represent grayscale (u16) format.\nHow do we know if it is 2D grayscale image or 2 separate layers in the mask representing two different forgeries?</li>\n</ol>",
  "messages": [
    {
      "id": "3381587",
      "postDate": "12/25/2025 01:22:52",
      "content": "<p>Any one know how to identify the below </p>\n<ol>\n<li>3D mask (3,h,w) layers when there are 3 different forged cells (one for each layer,) vs RGB data?\nThe numpy array header data is not useful as far as I know:\ne.g: {'descr': '|u1', 'fortran_order': False, 'shape': (3, 649, 884), }</li>\n<li>2D Numpy arrays (2,h,w) represent grayscale (u16) format.\nHow do we know if it is 2D grayscale image or 2 separate layers in the mask representing two different forgeries?</li>\n</ol>",
      "rawMarkdown": "Any one know how to identify the below \n1. 3D mask (3,h,w) layers when there are 3 different forged cells (one for each layer,) vs RGB data?\nThe numpy array header data is not useful as far as I know:\ne.g: {'descr': '|u1', 'fortran_order': False, 'shape': (3, 649, 884), }\n\n\n\n1. 2D Numpy arrays (2,h,w) represent grayscale (u16) format.\nHow do we know if it is 2D grayscale image or 2 separate layers in the mask representing two different forgeries?",
      "votes": null
    },
    {
      "id": "3381728",
      "postDate": "12/25/2025 11:07:45",
      "content": "<p>Check <a href=\"https://www.kaggle.com/competitions/recodai-luc-scientific-image-forgery-detection/discussion/664236\" target=\"_blank\">https://www.kaggle.com/competitions/recodai-luc-scientific-image-forgery-detection/discussion/664236</a> it will explain it for you. The number of channels (C, H, W) in the mask always means the number of different forgeries. It has nothing to do with RGB/grayscale.</p>",
      "rawMarkdown": "Check https://www.kaggle.com/competitions/recodai-luc-scientific-image-forgery-detection/discussion/664236 it will explain it for you. The number of channels (C, H, W) in the mask always means the number of different forgeries. It has nothing to do with RGB/grayscale.",
      "votes": null
    },
    {
      "id": "3381955",
      "postDate": "12/26/2025 03:34:08",
      "content": "<p>Thank you for clarifying.</p>",
      "rawMarkdown": "Thank you for clarifying.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3381728,
      "author_name": "returnofsputnik",
      "author_url": "",
      "post_date": "12/25/2025 11:07:45",
      "content": "<p>Check <a href=\"https://www.kaggle.com/competitions/recodai-luc-scientific-image-forgery-detection/discussion/664236\" target=\"_blank\">https://www.kaggle.com/competitions/recodai-luc-scientific-image-forgery-detection/discussion/664236</a> it will explain it for you. The number of channels (C, H, W) in the mask always means the number of different forgeries. It has nothing to do with RGB/grayscale.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3381955,
          "author_name": "venkatagoli",
          "author_url": "",
          "post_date": "12/26/2025 03:34:08",
          "content": "<p>Thank you for clarifying.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3381587": "Any one know how to identify the below \n1. 3D mask (3,h,w) layers when there are 3 different forged cells (one for each layer,) vs RGB data?\nThe numpy array header data is not useful as far as I know:\ne.g: {'descr': '|u1', 'fortran_order': False, 'shape': (3, 649, 884), }\n\n\n\n1. 2D Numpy arrays (2,h,w) represent grayscale (u16) format.\nHow do we know if it is 2D grayscale image or 2 separate layers in the mask representing two different forgeries?",
    "3381728": "Check https://www.kaggle.com/competitions/recodai-luc-scientific-image-forgery-detection/discussion/664236 it will explain it for you. The number of channels (C, H, W) in the mask always means the number of different forgeries. It has nothing to do with RGB/grayscale.",
    "3381955": "Thank you for clarifying."
  },
  "source": "meta"
}