NirAharon

Hi @jjjuurang, Did you try this code with your trained YOLOX? because if so, it is sound like compatibility issues between YOLOX versions.

@faziii0 did you change the Exp file self.num_classes from 1 and used trained multi-class YOLOX weights?

Hi @goldentimecoolk. I adjusted the TrackEval code to store the amount of FN, FP, IDSW and gtDet per frame and then calculated the cMOTA. It was pretty fast to calculate...

bot_sort.py is the original implementation I used to get the results in the paper. I didn't want to change this code because I wanted that people could reproduce our results...

Hello, The difference is that ByteTrack calculated the MOTA within their code while I used [TrackEval](https://github.com/JonathonLuiten/TrackEval) for evaluation - refer to [TrackEval/issues#75](https://github.com/JonathonLuiten/TrackEval/issues/75).

(1) 0.5 is because the cosine distance is in the range [0, 2] and the 1-IoU is in [0, 1]. (2) Yes, thank you. It is a typo in the...

I didn't try to evaluate on KITTI but I believe it should work fine. Please make sure to use the CMC we used in the paper and to train the...

Hi @mikel-brostrom, 1) Please don't use ORB or ECC for the CMC. Try to use the txt files we provided for the CMC or you can try the new sparse...

Hi @HanGuangXin, I used [TrackEval](https://github.com/JonathonLuiten/TrackEval) for getting the metrics from created text files. For evaluation the validation set, you can put [MOT17-valid.zip](https://github.com/NirAharon/BoT-SORT/files/9785990/MOT17-valid.zip) in TrackEval/data/gt/mot_challenge folder. Make sure to choose the...

Yes, you are right, I will fix this soon. Thank you.