I understand that YOLO is designed for object detection. What I was trying to do is perform a transfer learning using the Kinetic700 dataset over YOLOv5 for human activity recognition. Then I realize that the dataset doesn't have annotation for bounding box.
I didn't manage to find a annotation document for the dataset. And I am unable to do manual annotation due to time limitation.
Is there still a way to do transfer learning on YOLOv5 without the bounding box information?
Many thanks
The object detection model learns both from image features and bounding box coordinates. You wouldn't be able to do transfer learning without bounding boxes data.
The Kinetic700 dataset is an image classification dataset instead not object detection. That's why you can't find any bounding box data for it.
To address your problem, due to time limitations. You can find a similar public dataset to Kinetic700 that is actually purposed for object detection training. Or, you can find help from freelance annotators to annotate all the images manually.
