{
  "id": 417325,
  "title": "So, can the best solutions produce readable content on fragment 4?",
  "url": "/competitions/vesuvius-challenge-ink-detection/discussion/417325",
  "author_name": "",
  "post_date": "2023-06-15T08:54:00.159637500Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Now that the competition is over, it would be really awesome if we can see what's the <strong>visual</strong> result of some of the best models. In the end, that is of course why we have all been doing this!</p>\n<p>I am super curious to see whether the models can produce readable results on fragment 4.</p>\n<p><a href=\"https://www.kaggle.com/competition\" target=\"_blank\">@competition</a> hosts, would it be possible to share some screenshots of fragment 4 ground truth and best predictions?</p>",
  "messages": [
    {
      "id": "2303343",
      "postDate": "06/15/2023 08:54:00",
      "content": "<p>Now that the competition is over, it would be really awesome if we can see what's the <strong>visual</strong> result of some of the best models. In the end, that is of course why we have all been doing this!</p>\n<p>I am super curious to see whether the models can produce readable results on fragment 4.</p>\n<p><a href=\"https://www.kaggle.com/competition\" target=\"_blank\">@competition</a> hosts, would it be possible to share some screenshots of fragment 4 ground truth and best predictions?</p>",
      "rawMarkdown": "Now that the competition is over, it would be really awesome if we can see what's the **visual** result of some of the best models. In the end, that is of course why we have all been doing this!\n\nI am super curious to see whether the models can produce readable results on fragment 4.\n\n@competition hosts, would it be possible to share some screenshots of fragment 4 ground truth and best predictions?",
      "votes": null
    },
    {
      "id": "2303604",
      "postDate": "06/15/2023 11:29:08",
      "content": "<p>I guess it is not really readable. A score of 0.68, which currently is the best will mean many false positives and negatives and I doubt that this will result in readable text (for every 7 true positives you get ~2 false positives and ~8 false negatives if you balance FP and FN).</p>",
      "rawMarkdown": "I guess it is not really readable. A score of 0.68, which currently is the best will mean many false positives and negatives and I doubt that this will result in readable text (for every 7 true positives you get ~2 false positives and ~8 false negatives if you balance FP and FN).",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2303604,
      "author_name": "yankir",
      "author_url": "",
      "post_date": "06/15/2023 11:29:08",
      "content": "<p>I guess it is not really readable. A score of 0.68, which currently is the best will mean many false positives and negatives and I doubt that this will result in readable text (for every 7 true positives you get ~2 false positives and ~8 false negatives if you balance FP and FN).</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2303343": "Now that the competition is over, it would be really awesome if we can see what's the **visual** result of some of the best models. In the end, that is of course why we have all been doing this!\n\nI am super curious to see whether the models can produce readable results on fragment 4.\n\n@competition hosts, would it be possible to share some screenshots of fragment 4 ground truth and best predictions?",
    "2303604": "I guess it is not really readable. A score of 0.68, which currently is the best will mean many false positives and negatives and I doubt that this will result in readable text (for every 7 true positives you get ~2 false positives and ~8 false negatives if you balance FP and FN)."
  },
  "source": "meta"
}