Tim Meinhardt

`frame_id` should be continious within a sequence. Please make yourself familiar with the code, for example, the lines after this: https://github.com/timmeinhardt/trackformer/blob/e1dbc2536b86348c2f62c788c98a2faf06247147/src/generate_coco_from_mot.py#L183 In combination with the MOT GT files you can...

Please make yourself familiar with the code. Understand how this error can occur and then debug your indices. There is most definitely a simple explanation for this.

If you rename your dataset you must add it to the dataset factory. Please try to solve these issues on your own by making yourself familiar with the code and...

`train.json` and `val.json` is whats your problem. The tracking/inference part does not use the .json files used for training the model. For inference, we rely on the original MOT ground...

Your compilation throws a `filename longer than 260 characters` error which only happens on Windows (10) systems. I do not know how to resolve this and generally do not have...

You can try running a different CUDA version and install the PyTorch version closest to 1.5 working with that CUDA version. There is a good chance this will work. Check...

The default setting in [track.yaml](https://github.com/timmeinhardt/trackformer/blob/main/cfgs/track.yaml) is to run MOT17 train and test for all three sets of public detections `MOT17-ALL-ALL`. But you will only see metric evaluations for train cause...

Have you tried optimizing the tracking thresholds, for example, `detection_obj_score_thresh` and `track_obj_score_thresh`. Or visualizing the output? This should give you a good idea of what is wrong.

You could write a script to find the optimal hyperparameters. But there is no analytic way to find them. At first, I would visualize your outputs to understand whats happening....

1. At inference time, we do not match. The track queries transfer the identities implicitly. During training, the matching happens here: https://github.com/timmeinhardt/trackformer/blob/e468bf156b029869f6de1be358bc11cd1f517f3c/src/trackformer/models/detr_tracking.py#L81 2. The class loss represents different classes. IDs...