{
  "id": 578982,
  "title": "regarding 3d training and inference",
  "url": "/competitions/byu-locating-bacterial-flagellar-motors-2025/discussion/578982",
  "author_name": "",
  "post_date": "2025-05-14T13:53:15.629133Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi!</p>\n<p>i was really interested in this comp in the first month but i had to leave  for about 1.5 months because i was busy with some other local competitions and with my uni </p>\n<p>so i have some questions that will help me and many others if you don't mind</p>\n<p>1)  Is there enough  inference time for 3d models? (3d yolo, 3d DETR, 3d Unet)?</p>\n<p>2) if we had the enough inference time would they defeat the current 2d yolo approaches?</p>\n<p>3) Is there any new extra data other than the pinned one?</p>\n<p>4)did the cv/lb correlation  problem got fixed?</p>",
  "messages": [
    {
      "id": "3201860",
      "postDate": "05/14/2025 13:53:15",
      "content": "<p>Hi!</p>\n<p>i was really interested in this comp in the first month but i had to leave  for about 1.5 months because i was busy with some other local competitions and with my uni </p>\n<p>so i have some questions that will help me and many others if you don't mind</p>\n<p>1)  Is there enough  inference time for 3d models? (3d yolo, 3d DETR, 3d Unet)?</p>\n<p>2) if we had the enough inference time would they defeat the current 2d yolo approaches?</p>\n<p>3) Is there any new extra data other than the pinned one?</p>\n<p>4)did the cv/lb correlation  problem got fixed?</p>",
      "rawMarkdown": "Hi!\n\ni was really interested in this comp in the first month but i had to leave  for about 1.5 months because i was busy with some other local competitions and with my uni \n\nso i have some questions that will help me and many others if you don't mind\n\n\n1)  Is there enough  inference time for 3d models? (3d yolo, 3d DETR, 3d Unet)?\n\n2) if we had the enough inference time would they defeat the current 2d yolo approaches?\n\n3) Is there any new extra data other than the pinned one?\n\n4)did the cv/lb correlation  problem got fixed?",
      "votes": null
    },
    {
      "id": "3201922",
      "postDate": "05/14/2025 15:47:29",
      "content": "<p>3d inference is possible, I have to reduce image size though. I'm not beating the 2d models but I'm sure others will.</p>",
      "rawMarkdown": "3d inference is possible, I have to reduce image size though. I'm not beating the 2d models but I'm sure others will.",
      "votes": null
    },
    {
      "id": "3202034",
      "postDate": "05/14/2025 18:33:28",
      "content": "<p>how much score you got for 3d is it yolo or detr?</p>",
      "rawMarkdown": "how much score you got for 3d is it yolo or detr?",
      "votes": null
    },
    {
      "id": "3206112",
      "postDate": "05/20/2025 20:42:42",
      "content": "<p>Which option did you end up going with as far as 2D vs 2.5D vs 3D?  I have had success with both but have continued with 2D due to the processing speed and our last inference set.</p>",
      "rawMarkdown": "Which option did you end up going with as far as 2D vs 2.5D vs 3D?  I have had success with both but have continued with 2D due to the processing speed and our last inference set.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3201922,
      "author_name": "tennogh",
      "author_url": "",
      "post_date": "05/14/2025 15:47:29",
      "content": "<p>3d inference is possible, I have to reduce image size though. I'm not beating the 2d models but I'm sure others will.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3202034,
          "author_name": "tahaalshatiri",
          "author_url": "",
          "post_date": "05/14/2025 18:33:28",
          "content": "<p>how much score you got for 3d is it yolo or detr?</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 3206112,
      "author_name": "connorjd",
      "author_url": "",
      "post_date": "05/20/2025 20:42:42",
      "content": "<p>Which option did you end up going with as far as 2D vs 2.5D vs 3D?  I have had success with both but have continued with 2D due to the processing speed and our last inference set.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3201860": "Hi!\n\ni was really interested in this comp in the first month but i had to leave  for about 1.5 months because i was busy with some other local competitions and with my uni \n\nso i have some questions that will help me and many others if you don't mind\n\n\n1)  Is there enough  inference time for 3d models? (3d yolo, 3d DETR, 3d Unet)?\n\n2) if we had the enough inference time would they defeat the current 2d yolo approaches?\n\n3) Is there any new extra data other than the pinned one?\n\n4)did the cv/lb correlation  problem got fixed?",
    "3201922": "3d inference is possible, I have to reduce image size though. I'm not beating the 2d models but I'm sure others will.",
    "3202034": "how much score you got for 3d is it yolo or detr?",
    "3206112": "Which option did you end up going with as far as 2D vs 2.5D vs 3D?  I have had success with both but have continued with 2D due to the processing speed and our last inference set."
  },
  "source": "meta"
}