open-mmlab / mmtracking

OpenMMLab Video Perception Toolbox. It supports Video Object Detection (VID), Multiple Object Tracking (MOT), Single Object Tracking (SOT), Video Instance Segmentation (VIS) with a unified framework.
https://mmtracking.readthedocs.io/en/latest/
Apache License 2.0
3.58k stars 597 forks source link

The MOT tutorial does not output label for detection results #927

Open FCInter opened 11 months ago

FCInter commented 11 months ago

I'm running the tutorial. It works but I cannot get the predicted label for each box. I can only get the object_id, bbox, and confidence score in the tracking result, and only the bbox and confidence score in the detection result.

I also get the warning which says "Warning: The model doesn't have classes", just like what is shown in the tutorial.

I really need to get the predicted label for each detection. It's weird for a tracking model not providing the label.

How shall I fix this problem?

I'm looking forward to anyone's kind help. Thanks!!!