{
  "id": 301956,
  "title": "Whats your best model except the YOLO family(v5/x/r etc)?",
  "url": "/competitions/tensorflow-great-barrier-reef/discussion/301956",
  "author_name": "",
  "post_date": "2022-01-20T08:01:13.143218900Z",
  "votes": 6,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I am trying to work with effdets/fasterrcnns, but they are not even close to the YOLO models, don't know why.<br>\nWhich non YOLO model is best for you?</p>",
  "messages": [
    {
      "id": "1657471",
      "postDate": "01/20/2022 08:01:13",
      "content": "<p>I am trying to work with effdets/fasterrcnns, but they are not even close to the YOLO models, don't know why.<br>\nWhich non YOLO model is best for you?</p>",
      "rawMarkdown": "I am trying to work with effdets/fasterrcnns, but they are not even close to the YOLO models, don't know why.\nWhich non YOLO model is best for you?",
      "votes": null
    },
    {
      "id": "1657500",
      "postDate": "01/20/2022 08:32:13",
      "content": "<p>DetectoRS w/ ResNet 50 Backbone, at the moment sitting at 0.57 Public LB, currently retraining using Mosaic data augmentation and still also need to try the increased image size trick dominating the LB atm.</p>",
      "rawMarkdown": "DetectoRS w/ ResNet 50 Backbone, at the moment sitting at 0.57 Public LB, currently retraining using Mosaic data augmentation and still also need to try the increased image size trick dominating the LB atm.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1657500,
      "author_name": "maxvandijck",
      "author_url": "",
      "post_date": "01/20/2022 08:32:13",
      "content": "<p>DetectoRS w/ ResNet 50 Backbone, at the moment sitting at 0.57 Public LB, currently retraining using Mosaic data augmentation and still also need to try the increased image size trick dominating the LB atm.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1657471": "I am trying to work with effdets/fasterrcnns, but they are not even close to the YOLO models, don't know why.\nWhich non YOLO model is best for you?",
    "1657500": "DetectoRS w/ ResNet 50 Backbone, at the moment sitting at 0.57 Public LB, currently retraining using Mosaic data augmentation and still also need to try the increased image size trick dominating the LB atm."
  },
  "source": "meta"
}