Using Docker

Using Docker#

Serving algorithms in Docker containers can simplify their usage and deployment on different machines.

Start from the template#

To help you create a standard layout for your Imaging Server Kit project, you can use our cookiecutter template.

The template includes:

  • A README.md file to describe your project.

  • A requirements.txt file to list dependencies.

  • A recommended .gitignore file.

  • Docker files: Dockerfile, docker-compose.yml, and .dockerignore to build and run your algorithm server.

  • A main.py file where to implement your algorithm.

  • A samples/ folder where to include sample images.

Create a new project by running:

serverkit new <output_directory>

This generates a structure like:

serverkit-project
├── sample_images               # Sample images
│   └── blobs.tif
├── .gitignore
├── docker-compose.yml          # Run with `docker compose up`
├── Dockerfile
├── main.py                     # Algorithm implementation
├── README.md
└── requirements.txt

You will be prompted for a project name, project URL, and Python version. After generating your project, implement your algorithm in main.py. Consider adjusting requirements.txt, Dockerfile, README.md and other files as needed.

Build and run with docker-compose#

To build a docker image image and run your algorithm as a web server in a docker container, use the command:

docker compose up

This will build the docker image, install your algorithm, start the server and make it available on http://localhost:8000.

Next steps#

Congratulations! You have seen all of the main features of Imaging Server Kit.

The last two sections of this documentation explore:

  • Data layers - A reference of all available data layers.

  • Examples - A gallery of example Server Kit algorithms for selected projects.

Contents