Using my own YOLOX model trained with official YOLOX repo, but demo_track.py outputs nothing

Question

Using my own YOLOX model trained with official YOLOX repo, but demo_track.py outputs nothing

keqizero opened this issue 7 months ago · 4 comments

Hi, thank you for your great work!

I have a question about the YOLOX model. I am trying to achieve better results by replacing the YOLOX model with my own, which was trained using official YOLOX repo. However, after changing the model, I don't see any detection nor tracking results at all. The saved txt contains nothing. I assume it's because of the different model preparation, but I couldn't fix it.

Do you have any idea what causes the problem?

Answer 1 · 2024-05-14T08:41:55.000Z

@keqizero same problem. How did you solve that?

Answer 2 · 2024-05-14T08:45:22.000Z

@keqizero same problem. How did you solve that?

No I haven't solved it

Answer 3 · 2024-05-14T08:49:15.000Z

@keqizero I find official YOLOX ends with .pth, while ByteTrack ends with .pth.tar. Evaluation detection results from official YOLOX and ByteTrack are different. Thus, I am confusing on using which detector

Answer 4 · 2024-05-14T09:00:37.000Z

@keqizero I find official YOLOX ends with .pth, while ByteTrack ends with .pth.tar. Evaluation detection results from official YOLOX and ByteTrack are different. Thus, I am confusing on using which detector

I think the saving format shouldn't be a problem. It's just a name. The problem is the difference of the data preparation or model usage. I haven't looked into the code for a long time. I hope you have better luck :)