Segmentation on Polyp
84.5mIoULAW
Evaluation results
| Method | Links | ||
|---|---|---|---|
| LAWSynthetic Data=+ LAW (ours), Training Framework=nnUNet2026.03 | 84.5 | 83.2 | |
| SimTxtSeg-w-TVHASupervision=Weakly-Supervised, Annotation type=text2024.06 | 82.47 | 88.24 | |
| PraNetSupervision=Fully-Supervised2024.06 | 81.32 | 87.3 | |
| Pseudo-L:TVCC+SAMSupervision=Weakly-Supervised2024.06 | 81.06 | 87.46 | |
| SimTxtSeg-w/o-CMASupervision=Weakly-Supervised2024.06 | 80.83 | 87.22 | |
| Ariadne’s ThreadSupervision=Fully-Supervised2024.06 | 80.65 | 87.14 | |
| SimTxtSeg-w/o-CASupervision=Weakly-Supervised, Annotation type=text2024.06 | 80.64 | 86.87 | |
| WeakPolypSupervision=Weakly-Supervised, Annotation type=box2024.06 | 79.4 | 85.61 | |
| BoxPolypSupervision=Weakly-Supervised, Annotation type=box+half anno.2024.06 | 79.11 | 86.86 | |
| ResUNetSupervision=Fully-Supervised2024.06 | 75.31 | 82.6 | |
| SimTxtSeg-w/o-TVHASupervision=Weakly-Supervised2024.06 | 74.92 | 83.15 | |
| Adaptive DistillationSynthetic Data=+ Adaptive Distillation, Training Framework=nnUNet2026.03 | 72.7 | 79.5 | |
| Real OnlySynthetic Data=Real Only, Training Framework=nnUNet2026.03 | 70.9 | 78.3 | |
| ControlNet (Uniform)Synthetic Data=+ ControlNet (Uniform), Training Framework=nnUNet2026.03 | 68.8 | 75.9 | |
| boxshrinkSupervision=Weakly-Supervised, Annotation type=box2024.06 | 64.22 | 78.21 | |
| S2MESupervision=Weakly-Supervised, Annotation type=scribble2024.06 | 49.62 | 66.33 |