hi, I am using this repo on my own tracker and I have some questions about the usage.
the ground truth MOT20 data gives track ids from 1, 2, 3, etc. In my custom tracker, I output track ids in a different format, eg, 2000, 1001, 1000, etc. As per the mot_challenge_2d_box.py script, it says that the preprocessing function calculates the number of gt and tracker detections and unique track ids and relabels them to make them contiguous. Does this mean that it does not matter that my track id is in a different format compared to the gt track ids?
my tracker is probably not performing so well, so there may be some frames where there is zero detections. in this case, is it ok to leave that frame empty in my .txt tracker output files? For example, if my tracker detects nothing in frame 2, then there will not be any line entry for frame 2.
hi, I am using this repo on my own tracker and I have some questions about the usage.