TransT tracker integration

[dschoerk]

as originally posted in https://github.com/openvinotoolkit/cvat/issues/4768

Hi, I've recently started to use CVAT to annotate videos of traffic scenes. Before coming across CVAT I briefly experimented with Supervisely and was very impressed by their semi-automatic video annotation with TransT . TransT is a single object tracking algorithm and was presented recently at CVPR21, a Pytorch implementation is available here. So I picked that up and integrated it into CVAT as a new tracking tool. If you find this useful I would like to open a pull request for this new feature.

a demo video showing the workflow and a comparison between TransT and SiamMask if you are interested: https://www.youtube.com/watch?v=_oT44BXMaJY

[nmanovic]

Hi @joaocaldeira ,

Could you please explain why it is bad? In general we don't delete volumes specially. Usually we "down" a container if it is necessary to update the tool or maintain its host. I would say in general case users want to have data on the next time docker-compose up is run. What is your use case?

[joaocaldeira]

More directly, my point was that listing docker-compose down is inconsistent with the sentence preceding it:

The command below will stop and remove containers, networks, volumes, and images created by up.

To answer your question: my use case here was that I added some videos to the tool for testing when I first installed it, and then wanted to clear all data and restart cvat before adding the videos I actually want annotated.

[nmanovic]

I see. We will update README.md file. We got the line directly from documentation and it describes the command with all options. In our case we probably need to write "The command below will stop and remove containers and networks created by up". The sentence isn't precise (external network isn't removed) but I believe it is sufficient for the case. Will it work?

[joaocaldeira]

Yes, that sounds good. Thanks!