{
  "id": 572794,
  "title": "Very similar scenes",
  "url": "/competitions/image-matching-challenge-2025/discussion/572794",
  "author_name": "",
  "post_date": "2025-04-11T10:48:01.562227800Z",
  "votes": 6,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Looks like scenes that are split into multiple are the hardest ones to cluster. I haven't thought anything about them yet. I wonder how many of them are in the private dataset.  </p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2F7e505553fd3842cdf1bf2f339b7ec337%2FScreenshot%202025-04-11%20at%2013.42.17.png?generation=1744368325173634&amp;alt=media\" alt=\"vineyard\"></p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2F4b1853e122590d68de30f793996db6b3%2FScreenshot%202025-04-11%20at%2013.44.52.png?generation=1744368359542037&amp;alt=media\" alt=\"stairs\"></p>",
  "messages": [
    {
      "id": "3176434",
      "postDate": "04/11/2025 10:48:01",
      "content": "<p>Looks like scenes that are split into multiple are the hardest ones to cluster. I haven't thought anything about them yet. I wonder how many of them are in the private dataset.  </p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2F7e505553fd3842cdf1bf2f339b7ec337%2FScreenshot%202025-04-11%20at%2013.42.17.png?generation=1744368325173634&amp;alt=media\" alt=\"vineyard\"></p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2F4b1853e122590d68de30f793996db6b3%2FScreenshot%202025-04-11%20at%2013.44.52.png?generation=1744368359542037&amp;alt=media\" alt=\"stairs\"></p>",
      "rawMarkdown": "Looks like scenes that are split into multiple are the hardest ones to cluster. I haven't thought anything about them yet. I wonder how many of them are in the private dataset.  \n\n![vineyard](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2F7e505553fd3842cdf1bf2f339b7ec337%2FScreenshot%202025-04-11%20at%2013.42.17.png?generation=1744368325173634&alt=media)\n\n![stairs](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2F4b1853e122590d68de30f793996db6b3%2FScreenshot%202025-04-11%20at%2013.44.52.png?generation=1744368359542037&alt=media)",
      "votes": null
    },
    {
      "id": "3176484",
      "postDate": "04/11/2025 12:02:02",
      "content": "<p>Such scenes are taken in the spatially separated locations. But yes, they do look similar</p>",
      "rawMarkdown": "Such scenes are taken in the spatially separated locations. But yes, they do look similar",
      "votes": null
    },
    {
      "id": "3176527",
      "postDate": "04/11/2025 12:43:34",
      "content": "<p>Details matter!</p>",
      "rawMarkdown": "Details matter!",
      "votes": null
    },
    {
      "id": "3178203",
      "postDate": "04/13/2025 21:02:45",
      "content": "<p>There are also similar images in several other folders, but this folder is really quite similar. From the image, it’s clear that the color and pixel density are very close to each other, making it nearly impossible to separate the data in this folder. This is also mentioned in the general description of the competition.</p>",
      "rawMarkdown": "There are also similar images in several other folders, but this folder is really quite similar. From the image, it’s clear that the color and pixel density are very close to each other, making it nearly impossible to separate the data in this folder. This is also mentioned in the general description of the competition.",
      "votes": null
    },
    {
      "id": "3185325",
      "postDate": "04/23/2025 06:58:16",
      "content": "<p>I think we can process it in stages and then fuse the results from three different approaches.</p>",
      "rawMarkdown": "I think we can process it in stages and then fuse the results from three different approaches.",
      "votes": null
    },
    {
      "id": "3206766",
      "postDate": "05/21/2025 18:35:43",
      "content": "<p>they look similar. Thanks</p>",
      "rawMarkdown": "they look similar. Thanks",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3176484,
      "author_name": "oldufo",
      "author_url": "",
      "post_date": "04/11/2025 12:02:02",
      "content": "<p>Such scenes are taken in the spatially separated locations. But yes, they do look similar</p>",
      "votes": null,
      "replies": [
        {
          "id": 3176527,
          "author_name": "fabiobellavia",
          "author_url": "",
          "post_date": "04/11/2025 12:43:34",
          "content": "<p>Details matter!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 3178203,
      "author_name": "ali78kabirzadeh",
      "author_url": "",
      "post_date": "04/13/2025 21:02:45",
      "content": "<p>There are also similar images in several other folders, but this folder is really quite similar. From the image, it’s clear that the color and pixel density are very close to each other, making it nearly impossible to separate the data in this folder. This is also mentioned in the general description of the competition.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3185325,
      "author_name": "deepseeks",
      "author_url": "",
      "post_date": "04/23/2025 06:58:16",
      "content": "<p>I think we can process it in stages and then fuse the results from three different approaches.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3206766,
      "author_name": "",
      "author_url": "",
      "post_date": "05/21/2025 18:35:43",
      "content": "<p>they look similar. Thanks</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3176434": "Looks like scenes that are split into multiple are the hardest ones to cluster. I haven't thought anything about them yet. I wonder how many of them are in the private dataset.  \n\n![vineyard](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2F7e505553fd3842cdf1bf2f339b7ec337%2FScreenshot%202025-04-11%20at%2013.42.17.png?generation=1744368325173634&alt=media)\n\n![stairs](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2F4b1853e122590d68de30f793996db6b3%2FScreenshot%202025-04-11%20at%2013.44.52.png?generation=1744368359542037&alt=media)",
    "3176484": "Such scenes are taken in the spatially separated locations. But yes, they do look similar",
    "3176527": "Details matter!",
    "3178203": "There are also similar images in several other folders, but this folder is really quite similar. From the image, it’s clear that the color and pixel density are very close to each other, making it nearly impossible to separate the data in this folder. This is also mentioned in the general description of the competition.",
    "3185325": "I think we can process it in stages and then fuse the results from three different approaches.",
    "3206766": "they look similar. Thanks"
  },
  "source": "meta"
}