ifzhang/ByteTrack

Best setup for REID

ahsteven opened this issue · 2 comments

I am a bit confused on the best setup with your model for REID. In the paper it says you used fairmot for MOT17 reid results. The track example calls byte_tracker.py which seems to only use IOU. I see there is a fairmot demo in the tutorial so I am wondering the best setup. My plan is to use the bytesort trained model with the tracker.py file found in the fairmot tutorial. Is this the way to get the best setup? Thanks!

I'm having a similar problem. I tried to use BYTETrackerArgs and changed the buffer to 1000 but still, the tracker_id gets lost
when an object goes behind something and reappears
What to do
thanks