einarolafsson/cross-channel-cell-from-hoechst
Cross-channel cell from hoechst dataset 3029 paired fields. images/ input channel, masks/ instance-labelled ground truth (same stem = same field). Train/test annotation: fields.csv gives name,split,n_objects for every field - 2578 train (237957 objects) / 451 test (45098 objects). The split is by well ( split_by_well.csv ), so no well contributes to both sides. Prepared with spaCR (PyPI | conda-forge). Used to train einarolafsson/cross-channel-cell-from-hoechst-cpsam.
Cross-channel cell from hoechst dataset
3029 paired fields. images/ input channel, masks/ instance-labelled ground truth (same stem = same field).
Train/test annotation: fields.csv gives name,split,n_objects for every field - 2578 train (237957 objects) / 451 test (45098 objects). The split is by well ( split_by_well.csv ), so no well contributes to both sides.
Prepared with spaCR (PyPI | conda-forge). Used to train einarolafsson/cross-channel-cell-from-hoechst-cpsam.
