No ground truth for MOT20-04, skipping. about bytetrack

[Four1996]

I'm sorry but I don't really know about how the MOT Challenge submission work, but yes you have to submit the .txt result of the training set and you will only get the performance result on the test data only.

Thank you for your reply! I have a rough idea of the author's code flow: When I train, I need to make sure that lines 19 and 102 in yolo_x_mix_det are "train". For test, I need to make sure that lines 19 and 102 in yolo_x_mix_det are "train" in addition to running "track.py". So you get 21 of the.txt of the training set; When I need to train the test set, I need to modify line 19 and line 102 are "test" to get 21.txt belonging to the testing set. It can then be submitted to the MOT Challenge website. (I guess so)

[ifzhang]

It seems that you are running the results of the test set of MOT20 (MOT20-04, MOT20-06, MOT20-07, MOT20-08) and there are no GT files for test set. You can change line 19 and 102 in https://github.com/ifzhang/ByteTrack/blob/main/exps/example/mot/yolox_x_mix_mot20_ch.py to get the results of the training set of MOT20 (MOT20-01, MOT20-02, MOT20-03, MOT20-05)

[lbq779660843]

It seems that you are running the results of the test set of MOT20 (MOT20-04, MOT20-06, MOT20-07, MOT20-08) and there are no GT files for test set. You can change line 19 and 102 in https://github.com/ifzhang/ByteTrack/blob/main/exps/example/mot/yolox_x_mix_mot20_ch.py to get the results of the training set of MOT20 (MOT20-01, MOT20-02, MOT20-03, MOT20-05)

I change the code as you said above and it returned normal messages, but I still want to know to use test dataset just like 'We use the input size 1600 x 896 for MOT20-04, MOT20-07 and 1920 x 736 for MOT20-06, MOT20-08', which is written in README.md.

[Malikanhar]

Hi @lbq779660843, If you want to get the result on the test set, you have to submit the tracking result (which is stored in a .txt files) to the MOTChallenge website. I just submitted the ByteTrack tracking results on the MOTChallenge website and finally get the result on the test set.

[Four1996]

Hi @lbq779660843, If you want to get the result on the test set, you have to submit the tracking result (which is stored in a .txt files) to the MOTChallenge website. I just submitted the ByteTrack tracking results on the MOTChallenge website and finally get the result on the test set.

Hello, I have a question want to ask, I now take model trained MOT17 training set, then the test set is forecasted and predicted results want to commit to MOT Challenge the official web site, but met them at the time of submission requirements must be submitted with the training set and testing set all the TXT (21 training set and 21 testing set). However, I only got the predicted results of the test set (21). How did the MOT Challenge result submission work? If I had to predict the training set again, would that result in too good a result? (BECAUSE I did my training with MOT17)

[Malikanhar]

I'm sorry but I don't really know about how the MOT Challenge submission work, but yes you have to submit the .txt result of the training set and you will only get the performance result on the test data only.

[sheebarazzaq]

When we need to train the test set, I need to modify line 19 and line 101 are "test" to get 21.txt belonging to the testing set. It can then be submitted to the MOT Challenge website.
Line#19 - self.val_ann = "test.json" # change to train.json when running on training set
Line#101 - name='test', # change to train when running on training set

[sheebarazzaq]

Hello everyone, i have a question. When i got the 21 text files for test set but when i run for train set i got only 7 files text files? Why? Does anyone know how to get 21 files for train set?