List your company

Critical View of Safety assessment on Endoscapes (frame-level)

77.1Component 1 Score (C1)SurgAtlas-VLM
3.88422.89241.960.908Apr 24, 2026May 4, 2026May 14, 2026May 24, 2026Jun 3, 2026Jun 13, 2026Jun 24, 2026

Evaluation results

MethodLinks
SurgAtlas-VLMProtocol=Open-vocabulary generation2026.0677.182.97377.7
SurgVLM-72BProtocol=Open-vocabulary generation2026.0676.183.170.676.6
SurgVLM-7BProtocol=Open-vocabulary generation2026.0675.382.472.976.9
Qwen2.5-VL-7BProtocol=MCQ2026.0656.182.459.265.9
Gemini 2.0 FlashProtocol=MCQ2026.0647.863.967.159.6
InternVL3-8BProtocol=MCQ2026.064053.351.448.2
Sum-of-ChecksModel=GPT-4o mini2026.0433.629.147.936.9
Sum-of-ChecksModel=Claude Opus 3.52026.0431.132.946.937
Sum-of-ChecksModel=Claude Haiku 3.52026.0429.224.248.534
Direct+FSModel=Claude Opus 3.52026.042826.743.932.9
CoT+FSModel=Claude Opus 3.52026.0427.62643.432.3
Direct+FSModel=GPT-4o mini2026.0426.926.842.932.2
CoT+FSModel=GPT-4o mini2026.0426.928.244.333.1
CoTModel=Claude Opus 3.52026.0426.426.642.531.9
CoTModel=GPT-4o mini2026.0426.227.743.932.6
DirectModel=Claude Opus 3.52026.0424.325.442.930.9
SubQModel=Claude Opus 3.52026.0424.124.54330.5
SubQModel=Claude Haiku 3.52026.0423.422.543.329.8
DirectModel=GPT-4o mini2026.0422.923.942.529.7
SubQModel=GPT-4o mini2026.0422.525.943.430.6
CoT+FSModel=Claude Haiku 3.52026.0422.523.743.129.7
SubQ+FSModel=Claude Haiku 3.52026.0422.423.842.529.6
SubQ+FSModel=GPT-4o mini2026.0422.121.543.829.1
Direct+FSModel=Claude Haiku 3.52026.0421.121.946.529.8
DirectModel=Claude Haiku 3.52026.0420.720.842.227.9
CoTModel=Claude Haiku 3.52026.0419.721.743.828.4
SubQ+FSModel=Claude Opus 3.52026.0419.219.743.927.6
GPT-4oProtocol=MCQ2026.066.75.97.56.7