{
  "id": 417260,
  "title": "56th silver solutions (shake up from 76 to 56)",
  "url": "/competitions/vesuvius-challenge-ink-detection/discussion/417260",
  "author_name": "",
  "post_date": "2023-06-15T00:27:38.268837700Z",
  "votes": 9,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Thanks to  Kaggle for hosting this interesting competition!!!!<br>\nVery enjoyable competition!</p>\n<p>This competition was diffucult for us…</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2930242%2Ff58f7ab7cc16c2344edda69146922703%2F2023-06-15%20092546.png?generation=1686788768153679&amp;alt=media\" alt=\"\"></p>\n<p>Thank you <a href=\"https://www.kaggle.com/hengck23\" target=\"_blank\">@hengck23</a> and <a href=\"https://www.kaggle.com/tanakar\" target=\"_blank\">@tanakar</a><br>\n<a href=\"https://www.kaggle.com/code/tanakar/2-5d-segmentaion-baseline-training\" target=\"_blank\">https://www.kaggle.com/code/tanakar/2-5d-segmentaion-baseline-training</a><br>\n<a href=\"https://www.kaggle.com/competitions/vesuvius-challenge-ink-detection/discussion/407972\" target=\"_blank\">https://www.kaggle.com/competitions/vesuvius-challenge-ink-detection/discussion/407972</a></p>\n<h1>not worked for me</h1>\n<ul>\n<li>otsu's binalization for threshold free prediction</li>\n<li>morphology</li>\n<li>multi model ensemble ( regnety 032)</li>\n<li>pretraining with IR image</li>\n<li>insert Residual Refinment Module (BASNet)</li>\n<li>multi veiw modlel (<a href=\"https://valeoai.github.io/blog/publications/mvrss/\" target=\"_blank\">https://valeoai.github.io/blog/publications/mvrss/</a>)</li>\n</ul>\n<h1>our highscore model (not submit): 0.638</h1>\n<ul>\n<li>train resolution: 224, 448, 672</li>\n<li>model: seresnext26t</li>\n<li>valid : fragment 1 </li>\n</ul>",
  "messages": [
    {
      "id": "2302902",
      "postDate": "06/15/2023 00:27:38",
      "content": "<p>Thanks to  Kaggle for hosting this interesting competition!!!!<br>\nVery enjoyable competition!</p>\n<p>This competition was diffucult for us…</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2930242%2Ff58f7ab7cc16c2344edda69146922703%2F2023-06-15%20092546.png?generation=1686788768153679&amp;alt=media\" alt=\"\"></p>\n<p>Thank you <a href=\"https://www.kaggle.com/hengck23\" target=\"_blank\">@hengck23</a> and <a href=\"https://www.kaggle.com/tanakar\" target=\"_blank\">@tanakar</a><br>\n<a href=\"https://www.kaggle.com/code/tanakar/2-5d-segmentaion-baseline-training\" target=\"_blank\">https://www.kaggle.com/code/tanakar/2-5d-segmentaion-baseline-training</a><br>\n<a href=\"https://www.kaggle.com/competitions/vesuvius-challenge-ink-detection/discussion/407972\" target=\"_blank\">https://www.kaggle.com/competitions/vesuvius-challenge-ink-detection/discussion/407972</a></p>\n<h1>not worked for me</h1>\n<ul>\n<li>otsu's binalization for threshold free prediction</li>\n<li>morphology</li>\n<li>multi model ensemble ( regnety 032)</li>\n<li>pretraining with IR image</li>\n<li>insert Residual Refinment Module (BASNet)</li>\n<li>multi veiw modlel (<a href=\"https://valeoai.github.io/blog/publications/mvrss/\" target=\"_blank\">https://valeoai.github.io/blog/publications/mvrss/</a>)</li>\n</ul>\n<h1>our highscore model (not submit): 0.638</h1>\n<ul>\n<li>train resolution: 224, 448, 672</li>\n<li>model: seresnext26t</li>\n<li>valid : fragment 1 </li>\n</ul>",
      "rawMarkdown": "Thanks to  Kaggle for hosting this interesting competition!!!!\nVery enjoyable competition!\n\nThis competition was diffucult for us...\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2930242%2Ff58f7ab7cc16c2344edda69146922703%2F2023-06-15%20092546.png?generation=1686788768153679&alt=media)\n\nThank you @hengck23 and @tanakar\nhttps://www.kaggle.com/code/tanakar/2-5d-segmentaion-baseline-training\nhttps://www.kaggle.com/competitions/vesuvius-challenge-ink-detection/discussion/407972\n\n# not worked for me\n* otsu's binalization for threshold free prediction\n* morphology\n* multi model ensemble ( regnety 032)\n* pretraining with IR image\n* insert Residual Refinment Module (BASNet)\n* multi veiw modlel (https://valeoai.github.io/blog/publications/mvrss/)\n\n# our highscore model (not submit): 0.638\n* train resolution: 224, 448, 672\n* model: seresnext26t\n* valid : fragment 1",
      "votes": null
    },
    {
      "id": "2303283",
      "postDate": "06/15/2023 07:55:33",
      "content": "<p>Are there any other details, such as channel selection？</p>",
      "rawMarkdown": "Are there any other details, such as channel selection？",
      "votes": null
    },
    {
      "id": "2303448",
      "postDate": "06/15/2023 09:47:42",
      "content": "<p>We used between 19~51. <br>\nSame as <a href=\"https://www.kaggle.com/tanakar\" target=\"_blank\">@tanakar</a> 's code.</p>\n<pre><code>    \n     =   // \n     = mid - CFG.in_chans // \n     = mid + CFG.in_chans // \n     = range(start, end)\n</code></pre>",
      "rawMarkdown": "We used between 19~51. \nSame as @tanakar 's code.\n\n```\n    # idxs = range(65)\n    mid =  65 // 2\n    start = mid - CFG.in_chans // 2\n    end = mid + CFG.in_chans // 2\n    idxs = range(start, end)\n```",
      "votes": null
    },
    {
      "id": "2303460",
      "postDate": "06/15/2023 09:53:53",
      "content": "<p>Thank you for your reply. Congratulations!</p>",
      "rawMarkdown": "Thank you for your reply. Congratulations!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2303283,
      "author_name": "wangxuc",
      "author_url": "",
      "post_date": "06/15/2023 07:55:33",
      "content": "<p>Are there any other details, such as channel selection？</p>",
      "votes": null,
      "replies": [
        {
          "id": 2303448,
          "author_name": "sugupoko",
          "author_url": "",
          "post_date": "06/15/2023 09:47:42",
          "content": "<p>We used between 19~51. <br>\nSame as <a href=\"https://www.kaggle.com/tanakar\" target=\"_blank\">@tanakar</a> 's code.</p>\n<pre><code>    \n     =   // \n     = mid - CFG.in_chans // \n     = mid + CFG.in_chans // \n     = range(start, end)\n</code></pre>",
          "votes": null,
          "replies": [
            {
              "id": 2303460,
              "author_name": "wangxuc",
              "author_url": "",
              "post_date": "06/15/2023 09:53:53",
              "content": "<p>Thank you for your reply. Congratulations!</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2302902": "Thanks to  Kaggle for hosting this interesting competition!!!!\nVery enjoyable competition!\n\nThis competition was diffucult for us...\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2930242%2Ff58f7ab7cc16c2344edda69146922703%2F2023-06-15%20092546.png?generation=1686788768153679&alt=media)\n\nThank you @hengck23 and @tanakar\nhttps://www.kaggle.com/code/tanakar/2-5d-segmentaion-baseline-training\nhttps://www.kaggle.com/competitions/vesuvius-challenge-ink-detection/discussion/407972\n\n# not worked for me\n* otsu's binalization for threshold free prediction\n* morphology\n* multi model ensemble ( regnety 032)\n* pretraining with IR image\n* insert Residual Refinment Module (BASNet)\n* multi veiw modlel (https://valeoai.github.io/blog/publications/mvrss/)\n\n# our highscore model (not submit): 0.638\n* train resolution: 224, 448, 672\n* model: seresnext26t\n* valid : fragment 1",
    "2303283": "Are there any other details, such as channel selection？",
    "2303448": "We used between 19~51. \nSame as @tanakar 's code.\n\n```\n    # idxs = range(65)\n    mid =  65 // 2\n    start = mid - CFG.in_chans // 2\n    end = mid + CFG.in_chans // 2\n    idxs = range(start, end)\n```",
    "2303460": "Thank you for your reply. Congratulations!"
  },
  "source": "meta"
}