{
  "id": 668259,
  "title": "Evaluating Image Classification into GEL, PLANT, and OTHER Categories.",
  "url": "/competitions/recodai-luc-scientific-image-forgery-detection/discussion/668259",
  "author_name": "",
  "post_date": "2026-01-15T23:06:34.918658700Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I classified the input images into three experimental result categories (GEL, PLANT, and OTHER), but the differences between them turned out to be only slight.\nI had expected that electrophoresis images, in particular, would show clearer distinctions because the migration distance is important.\nAlthough the effect was smaller than I had hoped, there was still a modest improvement, so I’m sharing the classification list here.\nI would appreciate any comments on why the differences were not more pronounced.</p>",
  "messages": [
    {
      "id": "3391910",
      "postDate": "01/15/2026 23:06:34",
      "content": "<p>I classified the input images into three experimental result categories (GEL, PLANT, and OTHER), but the differences between them turned out to be only slight.\nI had expected that electrophoresis images, in particular, would show clearer distinctions because the migration distance is important.\nAlthough the effect was smaller than I had hoped, there was still a modest improvement, so I’m sharing the classification list here.\nI would appreciate any comments on why the differences were not more pronounced.</p>",
      "rawMarkdown": "I classified the input images into three experimental result categories (GEL, PLANT, and OTHER), but the differences between them turned out to be only slight.\nI had expected that electrophoresis images, in particular, would show clearer distinctions because the migration distance is important.\nAlthough the effect was smaller than I had hoped, there was still a modest improvement, so I’m sharing the classification list here.\nI would appreciate any comments on why the differences were not more pronounced.",
      "votes": null
    },
    {
      "id": "3391930",
      "postDate": "01/16/2026 00:45:53",
      "content": "<p>I similarly created a model to detect gell (the one with an elliptical black object on a gray background image) and corn, and changed the subsequent processing parameters based on those detection results.\nIn my case, while not perfect, the accuracy improved somewhat.</p>",
      "rawMarkdown": "I similarly created a model to detect gell (the one with an elliptical black object on a gray background image) and corn, and changed the subsequent processing parameters based on those detection results.\nIn my case, while not perfect, the accuracy improved somewhat.",
      "votes": null
    },
    {
      "id": "3393386",
      "postDate": "01/18/2026 23:49:40",
      "content": "<p>Thank you for your comment. I’m glad to hear that you were able to improve accuracy using the same strategy. In my case, I might have needed to put more thought into the pre‑ and post‑processing steps</p>",
      "rawMarkdown": "Thank you for your comment. I’m glad to hear that you were able to improve accuracy using the same strategy. In my case, I might have needed to put more thought into the pre‑ and post‑processing steps",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3391930,
      "author_name": "npinpi",
      "author_url": "",
      "post_date": "01/16/2026 00:45:53",
      "content": "<p>I similarly created a model to detect gell (the one with an elliptical black object on a gray background image) and corn, and changed the subsequent processing parameters based on those detection results.\nIn my case, while not perfect, the accuracy improved somewhat.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3393386,
          "author_name": "masarusanada",
          "author_url": "",
          "post_date": "01/18/2026 23:49:40",
          "content": "<p>Thank you for your comment. I’m glad to hear that you were able to improve accuracy using the same strategy. In my case, I might have needed to put more thought into the pre‑ and post‑processing steps</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3391910": "I classified the input images into three experimental result categories (GEL, PLANT, and OTHER), but the differences between them turned out to be only slight.\nI had expected that electrophoresis images, in particular, would show clearer distinctions because the migration distance is important.\nAlthough the effect was smaller than I had hoped, there was still a modest improvement, so I’m sharing the classification list here.\nI would appreciate any comments on why the differences were not more pronounced.",
    "3391930": "I similarly created a model to detect gell (the one with an elliptical black object on a gray background image) and corn, and changed the subsequent processing parameters based on those detection results.\nIn my case, while not perfect, the accuracy improved somewhat.",
    "3393386": "Thank you for your comment. I’m glad to hear that you were able to improve accuracy using the same strategy. In my case, I might have needed to put more thought into the pre‑ and post‑processing steps"
  },
  "source": "meta"
}