{
  "id": 305557,
  "title": "Is there a way to beat the 0.65 F2 score on YOLOV5?",
  "url": "/competitions/tensorflow-great-barrier-reef/discussion/305557",
  "author_name": "",
  "post_date": "2022-02-05T20:04:04.202892400Z",
  "votes": 8,
  "comment_count": 6,
  "views": 0,
  "content": "<p>I tried all the yolov5 models with dozen image sizes. But still my models can't beat the 0.65 F2 score. Is there a way to beat 0.65 F2 score using yolov5?</p>\n<p>Thanks.</p>",
  "messages": [
    {
      "id": "1677557",
      "postDate": "02/05/2022 20:04:04",
      "content": "<p>I tried all the yolov5 models with dozen image sizes. But still my models can't beat the 0.65 F2 score. Is there a way to beat 0.65 F2 score using yolov5?</p>\n<p>Thanks.</p>",
      "rawMarkdown": "I tried all the yolov5 models with dozen image sizes. But still my models can't beat the 0.65 F2 score. Is there a way to beat 0.65 F2 score using yolov5?\n\nThanks.",
      "votes": null
    },
    {
      "id": "1678367",
      "postDate": "02/06/2022 13:45:45",
      "content": "<p>too many, depends on your hardware and your methods.</p>",
      "rawMarkdown": "too many, depends on your hardware and your methods.",
      "votes": null
    },
    {
      "id": "1678462",
      "postDate": "02/06/2022 14:43:12",
      "content": "<p>For sure yes - our solution 0.703 (LB) is yolov5 …. and we use Colab for training.</p>",
      "rawMarkdown": "For sure yes - our solution 0.703 (LB) is yolov5 .... and we use Colab for training.",
      "votes": null
    },
    {
      "id": "1678479",
      "postDate": "02/06/2022 14:58:17",
      "content": "<p>I see. What was your f2 score? </p>",
      "rawMarkdown": "I see. What was your f2 score?",
      "votes": null
    },
    {
      "id": "1686752",
      "postDate": "02/12/2022 10:15:30",
      "content": "<p>Yes , My currently score 0.686 is based o yolov5 , colab also for training <br>\nGood luck </p>",
      "rawMarkdown": "Yes , My currently score 0.686 is based o yolov5 , colab also for training \nGood luck",
      "votes": null
    },
    {
      "id": "1687655",
      "postDate": "02/13/2022 03:20:49",
      "content": "<p>My score is 0.692 on YOLOV5, hardware and methods is important.</p>",
      "rawMarkdown": "My score is 0.692 on YOLOV5, hardware and methods is important.",
      "votes": null
    },
    {
      "id": "1687659",
      "postDate": "02/13/2022 03:24:11",
      "content": "<p>me too.😟😟😟</p>",
      "rawMarkdown": "me too.😟😟😟",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1678367,
      "author_name": "dragonzhang",
      "author_url": "",
      "post_date": "02/06/2022 13:45:45",
      "content": "<p>too many, depends on your hardware and your methods.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1678462,
      "author_name": "remekkinas",
      "author_url": "",
      "post_date": "02/06/2022 14:43:12",
      "content": "<p>For sure yes - our solution 0.703 (LB) is yolov5 …. and we use Colab for training.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1678479,
          "author_name": "kfk42kfk",
          "author_url": "",
          "post_date": "02/06/2022 14:58:17",
          "content": "<p>I see. What was your f2 score? </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1686752,
      "author_name": "asalhi",
      "author_url": "",
      "post_date": "02/12/2022 10:15:30",
      "content": "<p>Yes , My currently score 0.686 is based o yolov5 , colab also for training <br>\nGood luck </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1687655,
      "author_name": "fxalll",
      "author_url": "",
      "post_date": "02/13/2022 03:20:49",
      "content": "<p>My score is 0.692 on YOLOV5, hardware and methods is important.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1687659,
      "author_name": "freshair1996",
      "author_url": "",
      "post_date": "02/13/2022 03:24:11",
      "content": "<p>me too.😟😟😟</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1677557": "I tried all the yolov5 models with dozen image sizes. But still my models can't beat the 0.65 F2 score. Is there a way to beat 0.65 F2 score using yolov5?\n\nThanks.",
    "1678367": "too many, depends on your hardware and your methods.",
    "1678462": "For sure yes - our solution 0.703 (LB) is yolov5 .... and we use Colab for training.",
    "1678479": "I see. What was your f2 score?",
    "1686752": "Yes , My currently score 0.686 is based o yolov5 , colab also for training \nGood luck",
    "1687655": "My score is 0.692 on YOLOV5, hardware and methods is important.",
    "1687659": "me too.😟😟😟"
  },
  "source": "meta"
}