medSAM model REST API service

Standalone FastAPI service that serves the medSAM model.

1. setup environement variables

refer to examples/ - you will need to create .env (and .env.docker if you are using Docker) and place them in the root of the project folder.

tip: copy-paste the existing example files, remove the .example at then end, and add the correct variables per your configuration

2a. set up and run the service

set up a dedicated virtual environment to run the service

# (install uv)
# curl -LsSf https://astral.sh/uv/install.sh | sh
# https://docs.astral.sh/uv/getting-started/installation/

uv python install 3.11

uv sync

install pre-commit git hook scripts

uv run pre-commit install

start the service

# development
uv run fastapi dev --host 127.0.0.1 --port 54003 service.py

# production
uv run uvicorn service:app --host 127.0.0.1 --port 54003
# OR
uv run fastapi run --host 127.0.0.1 --port 54003 service.py

2b. run in a Docker container

build the Docker image

docker build -t medsam-model .

run the Docker container service

docker compose up

manually request segmentation for an image

# e.g. local
IMAGE_PATH="minio/{bucket_name}/{image_filename}"
# e.g. docker
IMAGE_PATH="local/{image_filename}"

# curl command once IMAGE_PATH has been set - saves image in data/ folder
curl -X GET "http://127.0.0.1:54003/process/${IMAGE_PATH}" -H "accept: application/json"

service documentation: http://127.0.0.1:54003/docs

NOTE: The service is designed to run locally and currently doesn't incorporate authentication or other security features.