{
  "id": 586402,
  "title": "Need more details on evaluation Metric !",
  "url": "/competitions/etaav-eo-ir/discussion/586402",
  "author_name": "",
  "post_date": "2025-06-26T14:30:04.751564900Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>can you explain how you are calculating f1 score of bounding boxes?<br>\nI tried calculating from classification matrix of Yolo Validation and also from bbox IoU calculation, however both do not match to score on LB</p>",
  "messages": [
    {
      "id": "3233174",
      "postDate": "06/26/2025 14:30:04",
      "content": "<p>can you explain how you are calculating f1 score of bounding boxes?<br>\nI tried calculating from classification matrix of Yolo Validation and also from bbox IoU calculation, however both do not match to score on LB</p>",
      "rawMarkdown": "can you explain how you are calculating f1 score of bounding boxes?\nI tried calculating from classification matrix of Yolo Validation and also from bbox IoU calculation, however both do not match to score on LB",
      "votes": null
    },
    {
      "id": "3236639",
      "postDate": "06/30/2025 14:04:20",
      "content": "<p>The F1 scores on the leaderboard (Kaggle) are automatically computed by comparing the CSV predictions submitted by participants with the ground truth results generated by our model. Therefore, you may need to focus on improving your model's generalization ability to achieve better validation accuracy.</p>\n<p>Hope this clears up your doubt.</p>",
      "rawMarkdown": "The F1 scores on the leaderboard (Kaggle) are automatically computed by comparing the CSV predictions submitted by participants with the ground truth results generated by our model. Therefore, you may need to focus on improving your model's generalization ability to achieve better validation accuracy.\n\nHope this clears up your doubt.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3236639,
      "author_name": "airliisc",
      "author_url": "",
      "post_date": "06/30/2025 14:04:20",
      "content": "<p>The F1 scores on the leaderboard (Kaggle) are automatically computed by comparing the CSV predictions submitted by participants with the ground truth results generated by our model. Therefore, you may need to focus on improving your model's generalization ability to achieve better validation accuracy.</p>\n<p>Hope this clears up your doubt.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3233174": "can you explain how you are calculating f1 score of bounding boxes?\nI tried calculating from classification matrix of Yolo Validation and also from bbox IoU calculation, however both do not match to score on LB",
    "3236639": "The F1 scores on the leaderboard (Kaggle) are automatically computed by comparing the CSV predictions submitted by participants with the ground truth results generated by our model. Therefore, you may need to focus on improving your model's generalization ability to achieve better validation accuracy.\n\nHope this clears up your doubt."
  },
  "source": "meta"
}