{
  "id": 468175,
  "title": "Possible 2D Approaches?",
  "url": "/competitions/blood-vessel-segmentation/discussion/468175",
  "author_name": "",
  "post_date": "2024-01-15T16:37:00.848610300Z",
  "votes": 5,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi, as the images of different kidneys have different shapes, I wanted to first cover all the 2D approaches. I think these are the possible ones-</p>\n<p>1 - Tile the Images (and stitch the preds)<br>\n2 - Resize (Downsample images then upsample mask)<br>\n3 - No Resizing (With small padding for the image to be divisible by 32)</p>\n<p>I am experimenting with tiling currently as the images are pretty big and the masks are very small, so I thought this would be a good approach. <br>\nWhich approach is the most viable for this competition in your opinion? Are you guys also using tiling?</p>",
  "messages": [
    {
      "id": "2603195",
      "postDate": "01/15/2024 16:37:00",
      "content": "<p>Hi, as the images of different kidneys have different shapes, I wanted to first cover all the 2D approaches. I think these are the possible ones-</p>\n<p>1 - Tile the Images (and stitch the preds)<br>\n2 - Resize (Downsample images then upsample mask)<br>\n3 - No Resizing (With small padding for the image to be divisible by 32)</p>\n<p>I am experimenting with tiling currently as the images are pretty big and the masks are very small, so I thought this would be a good approach. <br>\nWhich approach is the most viable for this competition in your opinion? Are you guys also using tiling?</p>",
      "rawMarkdown": "Hi, as the images of different kidneys have different shapes, I wanted to first cover all the 2D approaches. I think these are the possible ones-\n\n1 - Tile the Images (and stitch the preds)\n2 - Resize (Downsample images then upsample mask)\n3 - No Resizing (With small padding for the image to be divisible by 32)\n\nI am experimenting with tiling currently as the images are pretty big and the masks are very small, so I thought this would be a good approach. \nWhich approach is the most viable for this competition in your opinion? Are you guys also using tiling?",
      "votes": null
    },
    {
      "id": "2609203",
      "postDate": "01/19/2024 10:02:44",
      "content": "<p>tiles is good. <br>\nDownsample is bad for me</p>",
      "rawMarkdown": "tiles is good. \nDownsample is bad for me",
      "votes": null
    },
    {
      "id": "2611113",
      "postDate": "01/20/2024 15:04:34",
      "content": "<p>thank you!</p>",
      "rawMarkdown": "thank you!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2609203,
      "author_name": "aleksandrkruchinin",
      "author_url": "",
      "post_date": "01/19/2024 10:02:44",
      "content": "<p>tiles is good. <br>\nDownsample is bad for me</p>",
      "votes": null,
      "replies": [
        {
          "id": 2611113,
          "author_name": "shashwatraman",
          "author_url": "",
          "post_date": "01/20/2024 15:04:34",
          "content": "<p>thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2603195": "Hi, as the images of different kidneys have different shapes, I wanted to first cover all the 2D approaches. I think these are the possible ones-\n\n1 - Tile the Images (and stitch the preds)\n2 - Resize (Downsample images then upsample mask)\n3 - No Resizing (With small padding for the image to be divisible by 32)\n\nI am experimenting with tiling currently as the images are pretty big and the masks are very small, so I thought this would be a good approach. \nWhich approach is the most viable for this competition in your opinion? Are you guys also using tiling?",
    "2609203": "tiles is good. \nDownsample is bad for me",
    "2611113": "thank you!"
  },
  "source": "meta"
}