--- license: cc-by-4.0 library_name: cellpose tags: - cellpose - cellpose-sam - organoid - segmentation - microscopy --- # organoid-segmenter (`shape_seg_co`) A Cellpose-SAM (`cpsam`) checkpoint fine-tuned for whole-organoid instance segmentation of human cardiac organoids in phase-contrast images. It is the organoid model of [SHAPE](https://github.com/xuekunwu/SHAPE)'s `Organoid_Segmenter_Tool` in the cardiac organoid case studies: - `case studies/cardiac organoids/timecourse` (`--organoid-model`), - `case studies/cardiac organoids/drug_screen` (`CO_ORGANOID_MODEL`), - `case studies/cardiac organoids/segmentation` (the `shape_seg_co` arm, `--model-path`). | file | size | sha256 | |---|---|---| | `shape_seg_co` | 1,218,642,694 bytes | `f125b9754521efa042ef1ee614d8d900843b2c933517f393daf2e37268d20c48` | The sha256 is the one the published SHAPE runs record (`organoid_model_sha256`). ## Download ``` hf download SHAPE-agent/organoid-segmenter shape_seg_co --local-dir models ``` ## Use With SHAPE, pass the file to the tool (`model_path`) or set `SHAPE_ORGANOID_MODEL_PATH=models/shape_seg_co`. With Cellpose 4 directly: ```python from cellpose import models model = models.CellposeModel(gpu=True, pretrained_model="models/shape_seg_co") masks, flows, styles = model.eval(image) ``` ## License CC BY 4.0.