Wenjie

Results 5 comments of Wenjie

Reproduced results on validation set MOT15 with the provided weights(motr_final_0915.pth), MOTA: 29.1% IDF1: 67.1%.

Hi, have you solved this problem? I got the same results with reproduced model.

Reproduced results on validation set MOT15 with the provided weights(motr_final_0915.pth), MOTA: 29.1% IDF1: 67.1%.

Thanks a lot. By the way, did you try training on MOT17+crowdhuman and testing on MOT17? Will this reproduce the results of the paper's comparison with SOTA experiment?

RuntimeError: Unknown layout in MultiScaleDeformableAttnFunction

The issue is similar to [issue](https://github.com/Vahe1994/AQLM/issues/20). Degrading torch to 2.1.0 works for me.

Inference speed on RTX 2080Ti Vs. RTX 3090

Thanks for the response. I am so curious and the performance boost is so huge. I wonder what is the primary difference between the initial code and the current one....

Can someone reproduce the results in the paper and provide some simple guidance?

We fixed some bugs and run the code. However, the CD we reproduced is quite different, much larger than that claimed in paper.