| Status |   |
| Package |   |
| Meta |    |
| Cite |  |
VisCy (blend of vision and cyto) is a deep learning pipeline for training and deploying computer vision models for image-based phenotyping at single-cell resolution.
VisCy is organized as a uv workspace monorepo:
| Package | Description | Install |
|---|---|---|
| viscy-data | Data loading and Lightning DataModules for microscopy | pip install viscy-data |
| viscy-models | Neural network architectures (UNet, contrastive, VAE) | pip install viscy-models |
| viscy-transforms | GPU-accelerated image transforms for microscopy | pip install viscy-transforms |
| viscy-utils | Shared ML infrastructure for microscopy | pip install viscy-utils |
| Application | Description | Install |
|---|---|---|
| Cytoland | Robust virtual staining of organelles from label-free images | uv pip install -e "applications/cytoland" |
| DynaCLR | Self-supervised contrastive learning for cellular dynamics | uv pip install -e "applications/dynaclr" |
Install individual packages (e.g.):
pip install viscy-modelsOr install from source with all development dependencies:
git clone https://github.com/mehta-lab/VisCy.git
cd VisCy
uv syncFull documentation is hosted at https://mehta-lab.github.io/VisCy/stable/.
See CONTRIBUTING.md for development setup and guidelines.