Surgical workflow recognition on M2CAI 2016
89.45AccuracyUniSurg
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| UniSurg2026.02 | 89.45 | 82.91 | 75.42 | |
| Dino v3Scale=L2026.02 | 85.55 | 74.56 | 65.39 | |
| GastroNet2026.02 | 85.53 | 73.09 | 63.9 | |
| Dino v3Scale=H2026.02 | 80.01 | 68.92 | 58.65 | |
| VideoMAEScale=L2026.02 | 77.24 | 65.72 | 54.55 | |
| PL-StitchPretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | 77.1 | — | — | |
| PL-StitchPretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 76.4 | 69.1 | — | |
| SurgVLP2026.02 | 75.85 | 61.81 | 50.49 | |
| iBOTType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 71 | 64.5 | — | |
| SurgeNetScale=XL2026.02 | 69.87 | 55.23 | 44.04 | |
| LemonFMPretraining data=LEMON, Evaluation protocol=linear probing2025.03 | 68.4 | 62.4 | — | |
| LemonFMPretraining data=LEMON, Backbone=ConvNeXt-B, Protocol=Linear probing2025.11 | 68.4 | 62.4 | — | |
| iBOTType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | 68 | — | — | |
| SurgeNetXLPretraining data=SurgeNetXL, Evaluation protocol=linear probing2025.03 | 67.8 | 61.9 | — | |
| DINOType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 67.6 | 62.8 | — | |
| LemonFMPretraining data=LEMON, Backbone=ConvNeXt-B, Protocol=k-NN2025.11 | 65.8 | — | — | |
| DINOType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | 65.2 | — | — | |
| T-CoReType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 61.8 | 57.1 | — | |
| VideoMAEScale=G2026.02 | 60.73 | 46.75 | 35.71 | |
| T-CoReType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | 60.2 | — | — | |
| Endo-Mamba2026.02 | 60 | 42.95 | 32.52 | |
| Endo SSL2026.02 | 54.22 | 33.62 | 24.75 | |
| Endo-FMPretraining data=Private data, Evaluation protocol=linear probing2025.03 | 53.8 | 49.8 | — | |
| Endo-FMType=S, Pretraining data=Private data, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 53.8 | 49.8 | — | |
| EndoFM2026.02 | 53.47 | 34.46 | 25.32 | |
| VideoMAEv2Pretraining data=LEMON, Evaluation protocol=linear probing2025.03 | 50.7 | 41.3 | — | |
| VideoMAEv2Type=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 50.7 | 41.3 | — | |
| EndoViT2026.02 | 47.32 | 27.68 | 19.61 | |
| VideoMAEv2Type=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | 45.8 | — | — | |
| Selfsupsurg2026.02 | 45.23 | 21.03 | 15 | |
| Endo-FMType=S, Pretraining data=Private data, Backbone=ViT-B/16, Protocol=k-NN2025.11 | 44.8 | — | — | |
| MAEPretraining data=LEMON, Evaluation protocol=linear probing2025.03 | 40.7 | 38.1 | — | |
| MAEType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 40.7 | 38.1 | — | |
| EndoViTPretraining data=Merged public data, Evaluation protocol=linear probing2025.03 | 38.5 | 37.2 | — | |
| EndoViTType=S, Pretraining data=Merged public data, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 38.5 | 37.2 | — | |
| MAEType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | 36.4 | — | — | |
| GSViT2026.02 | 36.02 | 8.52 | 6.35 | |
| EndoViTType=S, Pretraining data=Merged public data, Backbone=ViT-B/16, Protocol=k-NN2025.11 | 34.6 | — | — | |
| GSViTPretraining data=GenSurgery, Evaluation protocol=linear probing2025.03 | 22.4 | 18 | — |