Surgical Video Question Answering on REAL-Colon-VQA In-template
87.69BLEU-4SurgLQA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SurgLQAzero-shot=false2026.05 | 87.69 | 92.93 | 91.85 | 79.6 | |
| Qwen3VLzero-shot=false2026.05 | 82.4 | 88.75 | 89.1 | 72.35 | |
| VideoLLaMA3zero-shot=false2026.05 | 76.47 | 85.6 | 84.75 | 68.34 | |
| RL framework over Digital Twin Representations2026.06 | 75.42 | 85.1 | 86.35 | 71.05 | |
| InternVL3zero-shot=false2026.05 | 72.3 | 83.15 | 85.22 | 65.91 | |
| SurgViVQALLM=GPT-2, Modality=Video, Training Paradigm=Ours2025.11 | 71.98 | 82.85 | 84.11 | 62.8 | |
| SurgViVQAzero-shot=false2026.05 | 71.98 | 82.85 | 84.11 | 63.2 | |
| SurgViVQA2026.06 | 71.98 | 82.85 | 84.11 | 63.2 | |
| MedGemmazero-shot=false2026.05 | 69.85 | 80.92 | 82.4 | 60.78 | |
| InternVL3 + LoRALLM=-, Modality=Video, Training Paradigm=FT2025.11 | 64.88 | 75.15 | 76.42 | 59.07 | |
| PitVQALLM=GPT-2, Modality=Image, Training Paradigm=Base2025.11 | 64.55 | 78.48 | 79.99 | 54.13 | |
| PitVQAzero-shot=false2026.05 | 64.55 | 78.48 | 79.99 | 54.13 | |
| PitVQA2026.06 | 64.55 | 78.48 | 79.99 | 54.13 | |
| SurgViVQALLM=Qwen, Modality=Video, Training Paradigm=Ours2025.11 | 60.09 | 74.36 | 75.32 | 59.73 | |
| MedGemmaLLM=-, Modality=Image, Training Paradigm=Zero-Shot2025.11 | 18.09 | 36.92 | 31.86 | 49.47 | |
| MedGemmaParameters=4B2026.06 | 18.09 | 36.92 | 31.86 | 49.47 | |
| SurgicalGPTLLM=GPT-2, Modality=Image, Training Paradigm=Base2025.11 | 14.93 | 47.85 | 52.36 | 33.33 | |
| SurgicalGPTzero-shot=false2026.05 | 14.93 | 47.85 | 52.36 | 33.33 | |
| SurgicalGPT2026.06 | 14.93 | 47.85 | 52.36 | 33.33 | |
| Qwen3-VLParameters=8B2026.06 | 8.9 | 50.1 | 54.7 | 58.2 | |
| Surgical-LVLM2026.06 | 7.9 | 49.8 | 53.8 | 57.2 | |
| Qwen3VLzero-shot=true2026.05 | 7.31 | 40.34 | 44.87 | 45.4 | |
| InternVL3LLM=-, Modality=Video, Training Paradigm=Zero-Shot2025.11 | 7.13 | 43.9 | 54.31 | 68.67 | |
| InternVL3zero-shot=true2026.05 | 7.13 | 43.9 | 54.31 | 68.67 | |
| InternVL3Parameters=1B2026.06 | 7.13 | 43.9 | 54.31 | 68.67 | |
| Qwen3-VLParameters=4B2026.06 | 6.8 | 48.2 | 51.5 | 54 | |
| Qwen2.5LLM=-, Modality=Image, Training Paradigm=Zero-Shot2025.11 | 4.53 | 45.7 | 46.83 | 49.33 | |
| Qwen2.5-VLParameters=3B2026.06 | 4.53 | 45.7 | 46.83 | 49.33 | |
| VideoLLaMA3LLM=-, Modality=Video, Training Paradigm=Zero-Shot2025.11 | 2.74 | 33.01 | 42.22 | 17.6 | |
| VideoLLaMA3Parameters=2B2026.06 | 2.74 | 33.01 | 42.22 | 17.6 |