Video-Docker

Some custom scripts for live video interaction with docker

On the (docker) client

reading video from real and fake webcams
accessing webcams on the host
accessing keyboard events on the host
displaying a window on the host

On the Host

listening to an incoming video feed
simulating an incoming video feed
sending the incoming video feed to a fake webcam on the client

Setup

For using the virtual camera:

apt-get install v4l2loopback-dkms
apt-get install v4l2 so you can use v4l2-ctl for troubleshooting
make sure to LD_PRELOAD or compile OpenCV with v4l2convert.so so it can read video4linux format. For Ubuntu 16.04 it should live at: /usr/lib/x86_64-linux-gnu/libv4l/v4l2convert.so

For autostart / standalone / gallery mode

Make sure to add v4l2loopback to /etc/modules so it's always loaded.
make the scripts executable (by soft-linking them in your users ~/bin folder
Check the Desktop folder for links to drag into Ubuntu's .config/autostart folder

Other things to consider:

Enable autologin for your user
Change the energy settings so your screensaver never turns on

For using docker:

install docker (plus nvidia-docker)
install deeplearning docker image of your choice (including opencv)
extend the docker file by adding v4l2 support