Docker Container
A docker container for pybioclip is hosted at ghcr.io. This container has CPU support for Mac, Windows, and Linux. GPU(CUDA) support is only for Windows and Linux.
Volumes
In order to access files the docker container requires you to mount a volume.
The working and home directories in this container are both set to /home/bcuser
.
Minimally you need to mount a volume into this directory so pybioclip can read your images.
When running pybioclip the software will download various BioCLIP files into a /home/bcuser/.cache
subdirectory. If you want to store the .cache
folder in your home directory you will need to mount that directory (~/.cache
) into the container at /home/bcuser/.cache
.
Setup
The examples below require an image file named Ursus-arctos.jpeg in the current directory.
Mac/Linux CPU Usage
The following command will create predictions for the Ursus-arctos.jpeg
image in the current directory.
The command mounts the current directory into the container at /home/bcuser
.
The command mounts the ~/.cache
directory into the container to cache BioCLIP files in your home directory.
docker run --platform linux/amd64 \
-v $(pwd):/home/bcuser \
-v ~/.cache:/home/bcuser/.cache \
--rm ghcr.io/imageomics/pybioclip:1.1.0 \
bioclip predict Ursus-arctos.jpeg
Linux GPU Usage
The following command will create predictions using a GPU for the Ursus-arctos.jpeg
image in the current directory.
docker run --gpus all \
--platform linux/amd64 \
-v $(pwd):/home/bcuser \
-v ~/.cache:/home/bcuser/.cache \
--rm ghcr.io/imageomics/pybioclip:1.1.0 \
bioclip predict --device cuda Ursus-arctos.jpeg
Windows CPU Usage
The following command will create predictions for the Ursus-arctos.jpeg
image in the current directory.
Since this command does not mount /home/bcuser/.cache
in the container the .cache
directory will be created within the current directory.
docker run --rm -v %cd%:/home/bcuser ghcr.io/imageomics/pybioclip:1.0.0 bioclip predict Ursus-arctos.jpeg
Windows GPU Usage
The following command will create predictions using a GPU for the Ursus-arctos.jpeg
image in the current directory.
docker run --rm --gpus all -v %cd%:/home/bcuser ghcr.io/imageomics/pybioclip bioclip:1.0.0 predict --device cuda Ursus-arctos.jpeg