Medical Image Segmentation on DRIVE relabeled
82.8Vessel ScoreRetSAM
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| RetSAMModel Architecture=RetSAM, Evaluation Protocol=Finetune2026.01 | 82.8 | 97.2 | 91.1 | 46.9 | 23.8 | |
| RetSAMModel Architecture=RetSAM, Evaluation Protocol=Linear2026.01 | 79.3 | 95.7 | 90.4 | 35.8 | 17.5 | |
| SAM2-UNetModel Architecture=SAM2-UNet2026.01 | 65.9 | 93.4 | 78.9 | — | — | |
| SAM3Model Architecture=SAM32026.01 | 65.5 | 92.5 | 74.8 | 8.5 | 14 | |
| SAM2Model Architecture=SAM22026.01 | 64.4 | 91.4 | 72.6 | 4.3 | 10 | |
| RetFoundModel Architecture=RetFound2026.01 | 57.9 | 89.5 | 54 | — | — | |
| MedSAM2Model Architecture=MedSAM22026.01 | 46.2 | 84.5 | 51.5 | — | 0.1 | |
| MedSAMModel Architecture=MedSAM2026.01 | 12.9 | 82.8 | 50.5 | 0.2 | 0.1 | |
| Med-SAM-AdapterModel Architecture=Med-SAM-Adapter2026.01 | 4.7 | 76.2 | 36.7 | — | — |