{
  "id": 297930,
  "title": "Yolo V5 (Head layer)",
  "url": "/competitions/tensorflow-great-barrier-reef/discussion/297930",
  "author_name": "Balasubramaniam",
  "post_date": "2021-12-30T16:13:04.745000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hii Kaggle Forks!</p>\n<p>I am recently working on an Object detection Model in Computer Vision. You Only Live once(YOLO)is the best Object Detection model compare to other detection models. Ha! Currently in YOLO Version 5 in PyTorch is creating the massive object detector in the year 2021.</p>\n<p>In the YOLOV5 version, its has been various pre-trained models to train the model for object detection. But we have an option to change the frozen params in a model. GitHub URL: <a href=\"https://github.com/ultralytics/yolov5\" target=\"_blank\">https://github.com/ultralytics/yolov5</a></p>\n<p>But I need to change the Head layer with connecting the transfer learning models to perform the YOLO for object detection</p>\n<p>what I mean exactly is the Transfer Learning model + YOLOV5 backbone is it possible for implementing the model for Higher Accuracy compared with the default model. </p>\n<p>In my training, I got a less accurate Result for taking a large time to perform the training to 8 hours but it not improving in the model!</p>\n<p>A kagglers please suggest to me how to efficiently modify the model with transfer layers</p>\n<p>I think this is my stupid question but I need to clear my doubt:)</p>",
  "messages": [
    {
      "id": 1633277,
      "postDate": "2021-12-30T16:13:04.747Z",
      "content": "<p>Hii Kaggle Forks!</p>\n<p>I am recently working on an Object detection Model in Computer Vision. You Only Live once(YOLO)is the best Object Detection model compare to other detection models. Ha! Currently in YOLO Version 5 in PyTorch is creating the massive object detector in the year 2021.</p>\n<p>In the YOLOV5 version, its has been various pre-trained models to train the model for object detection. But we have an option to change the frozen params in a model. GitHub URL: <a href=\"https://github.com/ultralytics/yolov5\" target=\"_blank\">https://github.com/ultralytics/yolov5</a></p>\n<p>But I need to change the Head layer with connecting the transfer learning models to perform the YOLO for object detection</p>\n<p>what I mean exactly is the Transfer Learning model + YOLOV5 backbone is it possible for implementing the model for Higher Accuracy compared with the default model. </p>\n<p>In my training, I got a less accurate Result for taking a large time to perform the training to 8 hours but it not improving in the model!</p>\n<p>A kagglers please suggest to me how to efficiently modify the model with transfer layers</p>\n<p>I think this is my stupid question but I need to clear my doubt:)</p>",
      "rawMarkdown": "Hii Kaggle Forks!\n\nI am recently working on an Object detection Model in Computer Vision. You Only Live once(YOLO)is the best Object Detection model compare to other detection models. Ha! Currently in YOLO Version 5 in PyTorch is creating the massive object detector in the year 2021.\n\nIn the YOLOV5 version, its has been various pre-trained models to train the model for object detection. But we have an option to change the frozen params in a model. GitHub URL: https://github.com/ultralytics/yolov5\n\nBut I need to change the Head layer with connecting the transfer learning models to perform the YOLO for object detection\n\nwhat I mean exactly is the Transfer Learning model + YOLOV5 backbone is it possible for implementing the model for Higher Accuracy compared with the default model. \n\n\nIn my training, I got a less accurate Result for taking a large time to perform the training to 8 hours but it not improving in the model!\n \nA kagglers please suggest to me how to efficiently modify the model with transfer layers\n\nI think this is my stupid question but I need to clear my doubt:)",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1633277": "Hii Kaggle Forks!\n\nI am recently working on an Object detection Model in Computer Vision. You Only Live once(YOLO)is the best Object Detection model compare to other detection models. Ha! Currently in YOLO Version 5 in PyTorch is creating the massive object detector in the year 2021.\n\nIn the YOLOV5 version, its has been various pre-trained models to train the model for object detection. But we have an option to change the frozen params in a model. GitHub URL: https://github.com/ultralytics/yolov5\n\nBut I need to change the Head layer with connecting the transfer learning models to perform the YOLO for object detection\n\nwhat I mean exactly is the Transfer Learning model + YOLOV5 backbone is it possible for implementing the model for Higher Accuracy compared with the default model. \n\n\nIn my training, I got a less accurate Result for taking a large time to perform the training to 8 hours but it not improving in the model!\n \nA kagglers please suggest to me how to efficiently modify the model with transfer layers\n\nI think this is my stupid question but I need to clear my doubt:)"
  }
}