Docker based setup

[jeverling]

I don't know if it's even possible - there is the --device flag that allows a container to access devices like webcams: https://medium.com/@zwinny/docker-using-webcam-9fafb26cf1e6

But I'm not sure if it's possible for the host to access the virtual camera device if it's created inside a container.

But if possible, it would be cool to have this working as a Docker container, so you can just fire it up on a Linux based system without having to install anything into your host.

[jeverling]

Seems to be possible: https://elder.dev/posts/open-source-virtual-background/

[alievk]

@jeverling it would be great to have a Docker setup, indeed. However, not a critical issue at the moment.
Please let us know if you have some progress in this direction :)

[bijanmmarkes]

I think the hardest part about Dockerizing this specific application would be controlling the interactive keys, and also the aspect of having it access Zoom or Skype from within the container.

With Docker, it's best to share the devices with the container rather than adding it to the python code.
--devices with docker-compose should work efficiently, I'm giving this a go.

[Rohlik]

I looking forward to see Docker support.

[mintmaker]

I am also looking forward to Docker support. It's pretty nice to not have to install things like ROS, TensorFlow and so on because you can have different versions of everything and only have to update the docker image. At the moment I am trying to hold my new Ubuntu 20.04 clean and dockerized, so it would be really neat.

I'm a bit experimenting with it at the moment with regard to the post @jeverling linked.
@bijanmmarkes Do you made progress?

Edit: Stopped experimenting right now (but may proceed later) and started install pytorch locally. But I would use docker support as soon as possible.

[mikaelhg]

See #224.

[mintmaker]

#224 solves this issue, but you still have to create the virtual camera outside the container
Edit: So I guess this can be closed.

[mintmaker]

This can be closed now. #224 has been merged and supports running everything (server and client side, and running on the machine) dockerized.