Surgical workflow recognition on OphNet
73.04AccuracyUniSurg
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| UniSurg2026.02 | 73.04 | 51.22 | 43.83 | |
| DinoV3-HBackbone variant=Huge2026.02 | 63.31 | 38.93 | 31.68 | |
| DinoV3-LBackbone variant=Large2026.02 | 62.8 | 39.99 | 31.99 | |
| GastroNet2026.02 | 61.41 | 37.95 | 30.56 | |
| VideoMAE-LBackbone variant=Large2026.02 | 58.97 | 34.35 | 27 | |
| VideoMAE-GBackbone variant=Giant2026.02 | 51.53 | 27.98 | 21.38 | |
| SurgVLP2026.02 | 48.13 | 24.61 | 18.76 | |
| Endo Mamba2026.02 | 28.94 | 10.37 | 7.53 | |
| EndoSSL2026.02 | 28.33 | 9.73 | 6.86 | |
| EndoViT2026.02 | 28.14 | 9.57 | 6.91 | |
| EndoFM2026.02 | 27.88 | 8.06 | 5.66 | |
| SurgeNetXLBackbone variant=XL2026.02 | 24.97 | 7.69 | 5.79 | |
| Selfsupsurg2026.02 | 23.15 | 6.36 | 4.48 | |
| GSViT2026.02 | 17.81 | 2.52 | 1.69 |