Closed valentinitnelav closed 1 year ago
The weights used were documented in the README file for the cluster scripts at https://github.com/stark-t/PAI/blob/dev_cluster/scripts/cluster/readme.md
We used this specific URLs to download the weights:
# YOLOv5 - nano
wget https://github.com/ultralytics/yolov5/releases/download/v6.1/yolov5n.pt -P ~/PAI/detectors/yolov5/weights_v6_1/
# YOLOv5 - small
wget https://github.com/ultralytics/yolov5/releases/download/v6.1/yolov5s.pt -P ~/PAI/detectors/yolov5/weights_v6_1/
# YOLOv7 - tiny
wget https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-tiny.pt -P ~/PAI/detectors/yolov7/weights_v0_1/
Hi @stark-t ,
I just realised that YOLOv7 comes with an example in the readme file specifically for transfer learning weights - https://github.com/WongKinYiu/yolov7#transfer-learning
Do you know/have a guess how would this be different than using the usual weights from their table ? https://github.com/WongKinYiu/yolov7#performance
For YOLOv5, I used their weights provided here: https://github.com/ultralytics/yolov5/releases/tag/v6.1 section Assets. I used
wget
to download them one by one, e.g:For YOLOv7, I used: