Please find the comprehensive napari-based prototype of cellcanvas here: https://github.com/cellcanvas/cellcanvas-prototype
And please check back to find the full version of cellcanvas
For the Copick workflows, install pip install '.[server,client]' from this checkout.
Run python workflows/cellcanvas-server/server.py CONFIG.json, then
python workflows/napari-client/client.py --config_path CONFIG.json.
These legacy workflows use Zarr 2; the local napari-cellcanvas workflow uses Zarr 3.
Superpixel creation needs the optional cellcanvas-spp package.
For the headless Zarr 3 service, install pip install '.[service]' in a separate environment.
Run python -m cellcanvas_service CONFIG.json; "datasets" maps names to core DataSet.from_paths arguments (paths are relative to the working directory).
Use GET /datasets, POST /datasets/NAME/train and POST /datasets/NAME/predict with optional {"batch_size": 65536}.