Nuclei Instance Segmentation on CRAG Dpath (test)
0.785DiceCAPL-Net (Proposed)
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CAPL-Net (Proposed)full_model=true2022.07 | 0.785 | 0.519 | 0.681 | 0.769 | 0.524 | |
| Baseline+CA+PLcomponents=Category-Aware feature alignment + Pseudo-Labelling2022.07 | 0.781 | 0.517 | 0.675 | 0.772 | 0.522 | |
| Full-supervisedsupervision=Fully supervised on target images2022.07 | 0.778 | 0.526 | 0.683 | 0.783 | 0.535 | |
| Baseline+CAcomponents=Category-Aware feature alignment2022.07 | 0.772 | 0.502 | 0.661 | 0.773 | 0.51 | |
| Yang et al.adaptation_strategy=global-level feature alignment2022.07 | 0.766 | 0.494 | 0.648 | 0.765 | 0.496 | |
| Baselineadaptation=Unsupervised Domain Adaptation2022.07 | 0.75 | 0.455 | 0.604 | 0.759 | 0.458 | |
| PDAMadaptation_levels=pixel-level and feature-level2022.07 | 0.596 | 0.323 | 0.467 | 0.676 | 0.316 | |
| Source Onlyadaptation=None2022.07 | 0.378 | 0.201 | 0.336 | 0.768 | 0.259 |