napari-cellposesam

A napari plugin for cell segmentation with cellpose, including Cellpose-SAM (cellpose v4).

This is a fork of MouseLand/cellpose-napari, the original plugin by Carsen Stringer. It adds:

• support for cellpose v4 and the Cellpose-SAM (cpsam) models, while still working with cellpose v3
• z/channel axis selection for 3D and 4D images
• compatibility with new napari versions

The cellpose code was written by Carsen Stringer and Marius Pachitariu. To learn about Cellpose, read the paper or watch this talk.

For problems with this plugin, open an issue. For problems with cellpose itself, open an issue on the cellpose repo.

If you use this plugin, please cite the cellpose paper, and the Cellpose-SAM paper if you use the cpsam models (see the cellpose repo for the reference):

@article{stringer2021cellpose,
  title={Cellpose: a generalist algorithm for cellular segmentation},
  author={Stringer, Carsen and Wang, Tim and Michaelos, Michalis and Pachitariu, Marius},
  journal={Nature Methods},
  volume={18},
  number={1},
  pages={100--106},
  year={2021},
  publisher={Nature Publishing Group}
}

Installation

Requires Python 3.11 or newer. We recommend a fresh environment, e.g. with conda:

conda create -n napari-cellposesam python=3.12
conda activate napari-cellposesam

Install napari and the plugin with pip:

pip install "napari[all]" napari-cellposesam

Or install the plugin from the napari plugin manager (Plugins > Install/Uninstall Plugins).

To install the latest development version from GitHub:

pip install git+https://github.com/jeskowagner/napari-cellposesam

If you have issues installing cellpose, see the cellpose installation docs.

GPU

cellpose runs on CPU, NVIDIA GPUs (CUDA) and Apple silicon (MPS). For CUDA, install a CUDA build of torch following the PyTorch instructions before or after installing the plugin. The plugin uses the GPU automatically if torch can see one.

Optional: CPDINO models

The CPDINO models need DINOv3, which is not on PyPI and so is not installed automatically. The plugin offers to install it when you run segmentation with a CPDINO model, or install it yourself:

pip install git+https://github.com/facebookresearch/dinov3

Without it, cellpose logs a Could not import CPDINO warning on startup. This is harmless if you don't use those models.

Upgrading

pip install --upgrade napari-cellposesam cellpose

Usage

Open napari with the plugin docked:

napari -w napari-cellposesam

Sample data is available under File > Open Sample > napari-cellposesam.

Detailed usage is described in the documentation.

Contributing

Contributions are welcome. For development, clone the repo and install it in editable mode with the test dependencies:

pip install -e ".[testing]"
pytest

License

Distributed under the terms of the BSD-3 license, "napari-cellposesam" is free and open source software.

Dependencies

napari-cellposesam relies on:

• napari
• magicgui
• cellpose, which in turn uses torch, numpy, scipy, opencv and others