Open LifeIsSoSolong opened 5 years ago
Try to use this PyTorch implementation and (.pth -> weights -> .pth) coverter: https://github.com/ultralytics/yolov3#darknet-conversion
More about other recommended frameworks: https://github.com/AlexeyAB/darknet#yolo-v3-in-other-frameworks
Try to use this PyTorch implementation and (.pth -> weights -> .pth) coverter: https://github.com/ultralytics/yolov3#darknet-conversion
More about other recommended frameworks: https://github.com/AlexeyAB/darknet#yolo-v3-in-other-frameworks
thanks, I have tried project ultralytics before I open this issue, but it didn't work. Whatever , thank you very much . I will try to find other methods~
I had the same issue, I solved it by cloning to the latest repo of Alexy's.
I had the same issue, I solved it by cloning to the latest repo of Alexy's.
Thanks, your .pth file is trained from project ultralytics or project erik?
Please clone the latest repo of ultralytics and alexey's darknet. It's working.
1.I have trained yolov3 to detect person(only one class) in pytorch using project: https://github.com/eriklindernoren/PyTorch-YOLOv3
2.The trained paraments is .pth format, and I convert it to .weights(In up project, load .pth and save as .weights), then I use it in this darknet project detecting person but result is always None.
it seems that my converted weights has some bug. Who can give me some advice , thanks~