Tim Meinhardt

I am not familiar with onnx and hence can not give any support on this matter.

I experienced the same with missing proposals. The VID validation set has 176126 files and you provide 176126 proposals. However, the training set has 1122397 files but you only provide...

My `root_dir` is not writable. That is why I set the `WAND_DIR` to something else. However, your code checks write permission in root_dir. Then warns about it saying it will...

Below some exemplary Python output from my docker image. The warning is correct. My current directory `/workspace` is (intentionally) not writeable. But from my understanding that should not matter or...

@nate-wandb `os.access(os.getenv('WANDB_DIR') or ".", os.W_OK)` returns `True` for me which makes sense since `WANDB_DIR` is set. If I understood this correctly, [this](https://github.com/wandb/wandb/blob/5f70cd84bd9726a427e53cdbac32e632c644931b/wandb/sdk/wandb_settings.py#L83) function is being called with `root_dir=/workspace` and not...

While this might have worked in your case, removing all package version numbers is not a reliable way of moving an outdated codebase into the future.

The codbase works with up to Pytorch1.12.1. What ever CUDA version compatible with that PyTorch version should work. There might be multiple.

This looks like a version issue. Maybe try running `pip install wheel` in the environment.

Please make yourself familiar with the `src/generate_coco_from_mot.py` file. From there you should be able to figure out what `seq_length`, `frame_ID` and `first_frame_image_id` mean. The COCO annotation format is not designed...

The image ids should be continious, yes. How would you initialize them otherwise? You go through all images in the dataet and then give each image an id. These are...