Tim Meinhardt

How different is the outcome? In distributed training mode the trainings are not perfectly deterministic. Try running it on a single GPU to check if it is deterministic then.

During testing the code is deterministic but not during training. There is a small drift in loss values. The Deformable DETR codebase has the same drift and I think this...

Thank you for the PR. I will check and merge it into the repository. However, to avoid any unintended bugs your should not use a different torchvision version anyway. Stick...

Is `flashmap.mp4` your file or from any of the presented datasets? If the video sequence contains drastic camera changes or very low frame rates this indeed can pose a challenging...

Please make yourself familiar with the PyTorch `distributed` package. In `train.py` we call the `utils.init_distributed_mode` method which initialises distributed mode with `init_mode=args.dist_url` which in turn is set to `env://`. This...

Did you download the pretrained model files and unpack them in the correct directory?

You can set `resume=False` and it will not load any pretrained model files. Only the backbone will load an ImageNet petraining.

If you were able to reproduce our results by running the provided model file this issue is something you have to debug on your side. I would start with the...

Since you are loading the `mot17` config your command is correct if you want to train on a multi-object tracking dataset. But since this is not the case for you,...

Yes it is possible to perform validation on static data but not for tracking metrics. Try to use the Visdom visualization to see if your data is formatted correctly and...