detections are not associated into the track

[wshku]

I trained your models on my custom dataset and then ran the track.py with your track.yaml. The problem is that the detections are good in each frame, but these detections belonging to the same track are not associated, leading to a huge number of tracks. Could you give me some suggestions on how to set your track parameters in track.yaml? Appreciate.

[timmeinhardt]

Debugging on a custom dataset is beyond the support of this repository. But you should check out the classification scores and thresholds in track.yaml.