Medical Image Segmentation on Kvasir
95.7mDiceOBBSeg
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| OBBSegSupervision=OBB+Box2026.07 | 95.7 | — | — | — | |
| OBBSegSupervision=OBB+Circle2026.07 | 95.7 | — | — | — | |
| OBBSegSupervision=OBB+Scribble2026.07 | 94.2 | — | — | — | |
| SSFormerTraining images used=1450/25942024.03 | 92.6 | 87.4 | — | — | |
| TransUNetTraining images used=1450/25942024.03 | 91.3 | 85.7 | — | — | |
| OBBSegSupervision=OBB+Point2026.07 | 91.2 | — | — | — | |
| RollingUNetBackbone=CNN2025.12 | 90.78 | — | — | — | |
| IPS-16PNumber of Trainable Parameters=1.3M, Training images used=1450/2594, Prompt points=162024.03 | 90.2 | 83.5 | — | — | |
| WeakPolyp+M2OSupervision=OBB2026.07 | 90 | — | — | — | |
| PraNetTraining images used=1450/25942024.03 | 89.8 | 84 | — | — | |
| RollingUNet + MAPOBackbone=CNN2025.12 | 89.04 | — | — | — | |
| IPS-5PNumber of Trainable Parameters=1.3M, Training images used=1450/2594, Prompt points=52024.03 | 85.5 | 77.2 | — | — | |
| IPS-3P|16PNumber of Trainable Parameters=1.3M, Training images used=1450/2594, Training prompt points=3, Testing prompt points=162024.03 | 84.3 | 75.2 | — | — | |
| U2-Net + MAPOBackbone=CNN2025.12 | 84.09 | — | — | — | |
| Med2DNumber of Trainable Parameters=184.5M, Training images used=5838/7935, Prompt points=162024.03 | 83.2 | 74.8 | — | — | |
| Med2DNumber of Trainable Parameters=184.5M, Training images used=5838/7935, Prompt points=52024.03 | 82.2 | 73.5 | — | — | |
| Med2DNumber of Trainable Parameters=184.5M, Training images used=5838/7935, Prompt points=32024.03 | 82.1 | 73.5 | — | — | |
| IPS-3P|5PNumber of Trainable Parameters=1.3M, Training images used=1450/2594, Training prompt points=3, Testing prompt points=52024.03 | 82.1 | 73.2 | — | — | |
| U-netTraining images used=1450/25942024.03 | 81.8 | 74.6 | — | — | |
| TransAttUNet + MAPOBackbone=Hybrid2025.12 | 81.73 | — | — | — | |
| IPS-3PNumber of Trainable Parameters=1.3M, Training images used=1450/2594, Prompt points=32024.03 | 79.7 | 70.4 | — | — | |
| Medical SAM32025.10 | 79.16 | — | — | — | |
| MISSFormer + MAPOBackbone=ViT2025.12 | 78.57 | — | — | — | |
| GazeMedSegSupervision=Gaze2026.07 | 77.8 | — | — | — | |
| RobustSAMPrompt type=point prompts, Prompt count (K)=3, Inference mode=single-image, Input condition=degraded2026.04 | 77.5 | — | — | — | |
| RobustMedSAMPrompt type=point prompts, Prompt count (K)=3, Inference mode=single-image, Input condition=degraded2026.04 | 77.4 | — | — | — | |
| SAMPrompt type=point prompts, Prompt count (K)=3, Inference mode=single-image, Input condition=degraded2026.04 | 77 | — | — | — | |
| AGMMSupervision=Point2026.07 | 75.5 | — | — | — | |
| RobustMedSAM+SVDPrompt type=point prompts, Prompt count (K)=3, Inference mode=single-image, Input condition=degraded2026.04 | 75.4 | — | — | — | |
| U2-NetBackbone=CNN2025.12 | 75.22 | — | — | — | |
| SAMPrompt points=52024.03 | 75 | 64.5 | — | — | |
| U-Net + MAPOBackbone=CNN2025.12 | 74.08 | — | — | — | |
| MedAugmentModel=UNet2023.06 | 73.801 | 63.031 | — | 91 | |
| DuoAugmentModel=UNet2023.06 | 73.7 | 62.5 | — | 91.47 | |
| BoxTeacherSupervision=Box2026.07 | 73.3 | — | — | — | |
| H2Former + MAPOBackbone=Hybrid2025.12 | 73.28 | — | — | — | |
| TriAugmentModel=UNet2023.06 | 73.1 | 61.77 | — | 91.53 | |
| PointSupSupervision=Point2026.07 | 73 | — | — | — | |
| MonoAugmentModel=UNet2023.06 | 72.27 | 61.27 | — | 89.13 | |
| SAMPrompt points=162024.03 | 71.9 | 62 | — | — | |
| STDAModel=UNet2023.06 | 71.5 | 60.53 | — | 91 | |
| AGMMSupervision=Scribble2026.07 | 67.2 | — | — | — | |
| BoxInstSupervision=Box2026.07 | 65.7 | — | — | — | |
| U-NetBackbone=CNN2025.12 | 65.06 | — | — | — | |
| MedAugmentModel=SwinUNETR2023.06 | 60.67 | 47.5 | — | 87.03 | |
| SwinUNet + MAPOBackbone=ViT2025.12 | 59.83 | — | — | — | |
| SAMPrompt points=32024.03 | 58.9 | 47.1 | — | — | |
| MISSFormerBackbone=ViT2025.12 | 58.03 | — | — | — | |
| MonoAugmentModel=SwinUNETR2023.06 | 57.97 | 44.43 | — | 85.6 | |
| STDAModel=SwinUNETR2023.06 | 57.37 | 44.03 | — | 85.8 | |
| TriAugmentModel=SwinUNETR2023.06 | 54.37 | 40.53 | — | 83.67 | |
| TransAttUNetBackbone=Hybrid2025.12 | 53.8 | — | — | — | |
| SwinUNetBackbone=ViT2025.12 | 52.2 | — | — | — | |
| DuoAugmentModel=SwinUNETR2023.06 | 51.67 | 37.77 | — | 81.3 | |
| NoAugmentModel=SwinUNETR2023.06 | 50.93 | 37.07 | — | 80.27 | |
| H2FormerBackbone=Hybrid2025.12 | 50.9 | — | — | — | |
| NoAugmentModel=UNet2023.06 | 43.77 | 30.8 | — | 67.37 | |
| UniMedVL2025.10 | 35.55 | — | — | — | |
| MedSAMPrompt type=point prompts, Prompt count (K)=3, Inference mode=single-image, Input condition=degraded2026.04 | 19.5 | — | — | — | |
| SAM32025.10 | 0 | — | — | — | |
| MedSAM2026.01 | — | 85.8 | 69.2 | — | |
| SAM2026.01 | — | 72.5 | 47.6 | — | |
| SAM-Med2D2026.01 | — | 86.6 | 62.8 | — | |
| VNS-SAM2026.01 | — | 84.7 | 65.7 | — | |
| VNS-SAMEncoder Adapter=true2026.01 | — | 85.8 | 66.4 | — |