Surgical Phase Recognition on MultiBypass140
0.7418Phase-level PrecisionMIS-DINO
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| MIS-DINOBackbone=ViT-B, Fine-tuning=true, Pre-training protocol=self-supervised learning2026.02 | 0.7418 | — | — | 0.7195 | 0.8846 | 0.7705 | 0.6158 | |
| ZENBackbone=ViT-B, Fine-tuning=true, Pre-training protocol=existing pretrained2026.02 | 0.7381 | — | — | 0.7273 | 0.8942 | 0.7737 | 0.6236 | |
| SurgeNetBackbone=CAFormer18, Fine-tuning=true, Pre-training protocol=existing pretrained2026.02 | 0.7347 | — | — | 0.7278 | 0.8851 | 0.7714 | 0.6246 | |
| ZENBackbone=ViT-B2026.02 | 0.726 | — | — | 0.7268 | 0.8845 | 0.7697 | 0.6212 | |
| MIS-DINOv2Backbone=ViT-L2026.02 | 0.7256 | — | — | 0.7221 | 0.8815 | 0.7574 | 0.6156 | |
| SurgeNetBackbone=CAFormer182026.02 | 0.7175 | — | — | 0.7082 | 0.8812 | 0.7619 | 0.6096 | |
| MIS-DINOv2Backbone=ViT-B2026.02 | 0.7143 | — | — | 0.7043 | 0.874 | 0.7508 | 0.5973 | |
| MIS-MSNBackbone=ViT-B, Fine-tuning=true, Pre-training protocol=self-supervised learning2026.02 | 0.7099 | — | — | 0.7048 | 0.8833 | 0.7737 | 0.6047 | |
| MIS-DINOBackbone=ViT-B2026.02 | 0.7097 | — | — | 0.7076 | 0.8781 | 0.7592 | 0.5985 | |
| MIS-DINOv2Backbone=ViT-L, Fine-tuning=true, Pre-training protocol=self-supervised learning2026.02 | 0.7029 | — | — | 0.7038 | 0.8898 | 0.7831 | 0.6068 | |
| MIS-MoCoV3Backbone=ViT-B, Fine-tuning=true, Pre-training protocol=self-supervised learning2026.02 | 0.701 | — | — | 0.6889 | 0.8718 | 0.7346 | 0.582 | |
| ImageNetBackbone=ViT-B, Fine-tuning=true, Pre-training protocol=existing pretrained2026.02 | 0.6999 | — | — | 0.676 | 0.8672 | 0.723 | 0.569 | |
| MIS-MAEBackbone=ViT-B, Fine-tuning=true, Pre-training protocol=self-supervised learning2026.02 | 0.6991 | — | — | 0.6901 | 0.8707 | 0.7473 | 0.5829 | |
| MIS-DINOv2Backbone=ViT-B, Fine-tuning=true, Pre-training protocol=self-supervised learning2026.02 | 0.6958 | — | — | 0.6845 | 0.8744 | 0.7551 | 0.5805 | |
| PeskaVLPBackbone=ResNet50, Fine-tuning=true, Pre-training protocol=existing pretrained2026.02 | 0.6955 | — | — | 0.6821 | 0.8748 | 0.7521 | 0.5821 | |
| EndoFMBackbone=ViT-B, Fine-tuning=true, Pre-training protocol=existing pretrained2026.02 | 0.6891 | — | — | 0.6655 | 0.8587 | 0.7385 | 0.567 | |
| EndoViTBackbone=ViT-B, Fine-tuning=true, Pre-training protocol=existing pretrained2026.02 | 0.6887 | — | — | 0.6705 | 0.8581 | 0.7262 | 0.5631 | |
| MIS-MoCoV3Backbone=ViT-B2026.02 | 0.6779 | — | — | 0.661 | 0.8598 | 0.7351 | 0.5642 | |
| PeskaVLPBackbone=ResNet502026.02 | 0.6758 | — | — | 0.6648 | 0.8577 | 0.702 | 0.5615 | |
| MIS-MAEBackbone=ViT-B2026.02 | 0.6746 | — | — | 0.6614 | 0.8537 | 0.7133 | 0.5629 | |
| EndoFMBackbone=ViT-B2026.02 | 0.6561 | — | — | 0.6373 | 0.8403 | 0.6835 | 0.531 | |
| ImageNetBackbone=ViT-B2026.02 | 0.6531 | — | — | 0.6387 | 0.8425 | 0.694 | 0.5365 | |
| MIS-MSNBackbone=ViT-B2026.02 | 0.6333 | — | — | 0.6183 | 0.8316 | 0.6627 | 0.5134 | |
| EndoViTBackbone=ViT-B2026.02 | 0.616 | — | — | 0.5959 | 0.8185 | 0.6481 | 0.494 | |
| Surg-R1Thinking=true2026.03 | 0.5524 | — | 0.4994 | — | — | 0.5074 | 0.3202 | |
| GSViTBackbone=EfficientViT, Fine-tuning=true, Pre-training protocol=existing pretrained2026.02 | 0.5015 | — | — | 0.447 | 0.6835 | 0.4799 | 0.3404 | |
| GSViTBackbone=EfficientViT2026.02 | 0.4612 | — | — | 0.4053 | 0.6451 | 0.4296 | 0.3005 | |
| Gemini-3.0-Pro-ThinkingThinking=true2026.03 | 0.3174 | — | 0.2872 | — | — | 0.2919 | 0.1661 | |
| GPT-5.1-ThinkingThinking=true2026.03 | 0.27 | — | 0.2901 | — | — | 0.2932 | 0.1521 | |
| Llama-4-Scout-17B-16E-InstructThinking=false2026.03 | 0.1551 | — | 0.1192 | — | — | 0.1198 | 0.0514 | |
| Mistral-Small-3.1-24B-Instruct-2503Thinking=false2026.03 | 0.1083 | — | 0.1078 | — | — | 0.1044 | 0.0405 | |
| MiniCPM-V-2_6Thinking=false2026.03 | 0.0923 | — | 0.0908 | — | — | 0.0883 | 0.0406 | |
| InternVL3-8BThinking=false2026.03 | 0.0865 | — | 0.0829 | — | — | 0.0802 | 0.0246 | |
| MiniCPM-O-2_6Thinking=false2026.03 | 0.0814 | — | 0.0885 | — | — | 0.0894 | 0.0386 | |
| LLaVA-1.5-7BThinking=false2026.03 | 0.0766 | — | 0.143 | — | — | 0.0433 | 0.0209 | |
| Phi-4-MultimodalThinking=false2026.03 | 0.0753 | — | 0.0681 | — | — | 0.0676 | 0.0314 | |
| Gemma3-27B-itThinking=false2026.03 | 0.0681 | — | 0.0885 | — | — | 0.0867 | 0.0194 | |
| Qwen2.5-VL-7B-Surg-CholecT50Thinking=false2026.03 | 0.0546 | — | 0.0976 | — | — | 0.0978 | 0.0336 | |
| Qwen2.5-VL-7B-InstructThinking=false2026.03 | 0.0253 | — | 0.0829 | — | — | 0.0817 | 0.0172 | |
| Gemini 3.1 Pro2026.03 | — | 0.22 | — | — | — | — | — | |
| GPT-5.12026.03 | — | 0.13 | — | — | — | — | — | |
| Qwen3-VLParameters=8B2026.03 | — | 0.08 | — | — | — | — | — | |
| SureonVLM2026.03 | — | 0.4 | — | — | — | — | — |