Medseg Rai Pannuke
Trained weights (bestmodel.pth) for the MedSeg-RAI demo. U-Net++ with a ResNet-50 encoder, 6-class semantic segmentation of H&E nuclei: Background, Neoplastic, Inflammatory, Connective, Dead, Epithelial. Trained on...
Source du modèle
- Type
- Other
- Taille du fichier
- 187 MB
- Version
- vBest model
- Mis à jour
- 1 nov. 2026
- Source
- Hugging Face
Description de la source
MedSeg-RAI: PanNuke nucleus segmentation (U-Net++ ResNet-50)
Trained weights (best_model.pth) for the MedSeg-RAI demo. U-Net++ with a ResNet-50
encoder, 6-class semantic segmentation of H&E nuclei: Background, Neoplastic,
Inflammatory, Connective, Dead, Epithelial. Trained on PanNuke folds 1 and 2.
Test results on PanNuke fold 3 (with test-time augmentation): mean foreground Dice 0.644, robust mean 0.714, pixel accuracy 0.919.
The checkpoint is a torch dict containing the state_dict, the run config, and the class
names. Load it with the project's medseg.evaluate.load_run.
The model was trained on PanNuke (CC BY-NC-SA 4.0), so these weights inherit the same non-commercial, share-alike terms. Not a medical device; research and portfolio use only.
Live demo and source: see the Hugging Face Space and the GitHub repository.
Sources
1 source
Sources
1 sourceVérifié 1 nov.
Artefacts du modèle
1 artefact
Artefacts du modèle
1 artefactExtraits de sources
2 extraits
Extraits de sources
2 extraits--- license: cc-by-nc-sa-4.0 library_name: pytorch pipeline_tag: image-segmentation tags: - histopathology - medical-imaging - semantic-segmentation - pannuke - pytorch --- # MedSeg-RAI: PanNuke nucleus segmentation (U-Net++ ResNet-50) Trained weights (`best_model.pth`) for the MedSeg-RAI demo. U-Net++ with a ResNet-50 encoder, 6-class semantic segmentation of H&E nuclei: Background, Neoplastic, Inflammatory, Connective, Dead, Epithelial. Trained on PanNuke folds 1 and 2. Test results on PanNuke fold 3 (with test-time augmentation): mean foreground Dice 0.644, robust mean 0.714, pixel accuracy 0.919. The checkpoint is a torch dict containing the state_dict, the run config, and the class names. Load it with the project's `medseg.evaluate.load_run`. The model was trained on PanNuke (CC BY-NC-SA 4.0), so these weights inherit the same non-commercial, share-alike terms. Not a medical device; research and portfolio use only. Live demo and source: see the Hugging Face Space and the GitHub repository.
Source context: 0 downloads · 0 likes · Pipeline image-segmentation · Library pytorch · Repo Devisri515/medseg-rai-pannuke
