Surgical Phase Recognition on Autolaparo
69Average F1PL-Stitch
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PL-StitchPretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 69 | 79.9 | |
| LemonFMPretraining data=LEMON, Evaluation protocol=linear probing2025.03 | 66.9 | 76.4 | |
| iBOTType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 65.1 | 76.3 | |
| DINOType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 65 | 74.9 | |
| LemonFMPretraining data=LEMON, Backbone=ConvNeXt-B, Protocol=Linear probing2025.11 | 64.5 | 74.7 | |
| SPAShots=322025.06 | 60.36 | — | |
| SPAShots=162025.06 | 58.91 | — | |
| T-CoReType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 58 | 67.9 | |
| SurgeNetXLPretraining data=SurgeNetXL, Evaluation protocol=linear probing2025.03 | 57 | 68.8 | |
| SPAShots=12025.06 | 51.48 | — | |
| Linear probe (LP)Shots=322025.06 | 44.72 | — | |
| Endo-FMPretraining data=Private data, Evaluation protocol=linear probing2025.03 | 43.1 | 51.5 | |
| Endo-FMType=S, Pretraining data=Private data, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 43.1 | 51.5 | |
| Linear probe (LP)Shots=162025.06 | 42.6 | — | |
| VideoMAEv2Pretraining data=LEMON, Evaluation protocol=linear probing2025.03 | 42.4 | 49.8 | |
| VideoMAEv2Type=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 42.4 | 49.8 | |
| Tip-Adapter-FShots=322025.06 | 39.75 | — | |
| LP+textShots=162025.06 | 39.43 | — | |
| LP+textShots=322025.06 | 39.4 | — | |
| EndoViTPretraining data=Merged public data, Evaluation protocol=linear probing2025.03 | 38.7 | 45.4 | |
| EndoViTType=S, Pretraining data=Merged public data, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 38.7 | 45.4 | |
| Tip-Adapter-FShots=162025.06 | 37.51 | — | |
| Linear probe (LP)Shots=12025.06 | 33.9 | — | |
| LP+textShots=12025.06 | 32.8 | — | |
| MAEPretraining data=LEMON, Evaluation protocol=linear probing2025.03 | 32 | 35.5 | |
| MAEType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=Linear probing2025.11 | 32 | 35.5 | |
| PeskaVLPShots=02025.06 | 28.6 | — | |
| Tip-Adapter-FShots=12025.06 | 26.58 | — | |
| HecVLPretraining dataset=SVL-Pretrain, Zero-shot=true2024.05 | 18.9 | 23.3 | |
| CLIPShots=02025.06 | 18.3 | — | |
| GSViTPretraining data=GenSurgery, Evaluation protocol=linear probing2025.03 | 17.5 | 22 | |
| SurgVLPPretraining dataset=SVL-Pretrain, Zero-shot=true2024.05 | 16.6 | 21.3 | |
| CLIP*Pretraining dataset=SVL-Pretrain, Zero-shot=true, Initialization=OpenAI2024.05 | 16.2 | 18.9 | |
| SurgLIMEBackbone=ViT-B/16, Zero-shot=true2026.04 | 11.2 | 18.1 | |
| CLIP**Pretraining dataset=SVL-Pretrain, Zero-shot=true, Initialization=random2024.05 | 10.9 | 15.3 | |
| CLIPPretraining dataset=CLIP400M, Zero-shot=true2024.05 | 9.1 | 17.4 | |
| MIL-NCEPretraining dataset=Howto100M, Zero-shot=true2024.05 | 7.9 | 9.9 | |
| SurgVLPBackbone=ResNet50, Zero-shot=true2026.04 | 7.2 | 10 | |
| CLIPBackbone=ViT-B/16, Zero-shot=true2026.04 | 4.8 | 8 | |
| DINOType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | — | 72.8 | |
| Endo-FMType=S, Pretraining data=Private data, Backbone=ViT-B/16, Protocol=k-NN2025.11 | — | 46.8 | |
| EndoViTType=S, Pretraining data=Merged public data, Backbone=ViT-B/16, Protocol=k-NN2025.11 | — | 40.2 | |
| iBOTType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | — | 75.3 | |
| LemonFMPretraining data=LEMON, Backbone=ConvNeXt-B, Protocol=k-NN2025.11 | — | 73.1 | |
| MAEType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | — | 35.9 | |
| PL-StitchPretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | — | 82.5 | |
| T-CoReType=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | — | 61.8 | |
| VideoMAEv2Type=G, Pretraining data=LEMON, Backbone=ViT-B/16, Protocol=k-NN2025.11 | — | 46.8 |