Skip to content

docker.md

Latest commit

 

History

History
57 lines (37 loc) · 1.89 KB

docker.md

File metadata and controls

57 lines (37 loc) · 1.89 KB

Using DeepEthogram in Docker

DeepEthogram's Docker images are built with uv, so you do not need conda inside the container.

Install Docker: https://docs.docker.com/get-docker/

Install the NVIDIA Container Toolkit if you want GPU access: https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#docker

Running the GUI on Linux with training support

In a terminal, run xhost +local:docker. You'll need to do this each time you restart.

docker run --gpus all -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix:rw --shm-size 16G -v /media:/media -it jbohnslav/deepethogram:full

Explanation:

  • --gpus all: required to expose NVIDIA GPUs inside the container
  • -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix:rw: forwards your X11 display
  • --shm-size 16G: helps PyTorch multiprocessing workloads
  • -v /media:/media: mounts your data drive; replace /media with the path that makes sense on your machine

The full image launches deepethogram by default.

Running the GUI without training support

Again, replace /media with the path to your data.

docker run -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix:rw -v /media:/media -it jbohnslav/deepethogram:gui

Running tests in the headless image

docker run --gpus all -v /media:/media -it jbohnslav/deepethogram:headless pytest -v

Verifying the images

Use the repo-local images if you built them yourself:

docker run --gpus all -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix:rw -v /media:/media -it deepethogram:full
docker run -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix:rw -v /media:/media -it deepethogram:gui
docker run --gpus all -it deepethogram:headless pytest -v

Building it yourself

To build the full image from this repository:

docker buildx build --load --target full -t deepethogram:full -f docker/Dockerfile .