List your company

GUI Agent Robustness Evaluation on Pop-up Perturbation 1.0 (Large)

97.5DSR (%)CA
2.3427.04551.7576.455Jul 13, 2025

Evaluation results

MethodLinks
CABackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Overlay, Visual Prompting=Highlight2025.0797.51002.5
CABackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Overlay, Visual Prompting=Default2025.07971003
CABackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Inductive, Visual Prompting=Highlight2025.0791.599.58
CABackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Inductive, Visual Prompting=Default2025.07899910
NDBackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Overlay, Visual Prompting=Highlight2025.0775.480.95.5
NDBackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Overlay, Visual Prompting=Default2025.0770.9809.1
CABackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Overlay, Visual Prompting=Highlight2025.0770.986.916
CABackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Overlay, Visual Prompting=Default2025.0767.888.420.6
CABackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Inductive, Visual Prompting=Highlight2025.07678417
NDBackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Inductive, Visual Prompting=Highlight2025.0761.57715.5
CABackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Inductive, Visual Prompting=Default2025.0761.592.531
DPOBackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Overlay, Visual Prompting=Highlight2025.0757.376.919.6
DPOBackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Overlay, Visual Prompting=Default2025.0756.374.418.1
NDBackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Inductive, Visual Prompting=Default2025.07567317
DPOBackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Inductive, Visual Prompting=Highlight2025.0744.57227.5
DABackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Inductive, Visual Prompting=Highlight2025.07448137
DABackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Inductive, Visual Prompting=Default2025.0742.580.538
DABackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Inductive, Visual Prompting=Highlight2025.0742.582.540
DPOBackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Inductive, Visual Prompting=Default2025.074168.527.5
DABackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Inductive, Visual Prompting=Default2025.0740.582.542
DABackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Overlay, Visual Prompting=Highlight2025.0736.765.829.1
DABackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Overlay, Visual Prompting=Default2025.0736.265.829.6
DABackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Overlay, Visual Prompting=Highlight2025.0732.279.547.3
DABackbone=LLaVA-v1.6-Vicuna-13B (LaSM L12-28, α=1.2), Injection Type (IT)=Overlay, Visual Prompting=Default2025.0731.784.452.7
NDBackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Overlay, Visual Prompting=Highlight2025.0714.164.350.2
NDBackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Overlay, Visual Prompting=Default2025.0713.162.349.2
DPOBackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Overlay, Visual Prompting=Highlight2025.0711.663.852.2
NDBackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Inductive, Visual Prompting=Highlight2025.0710.571.561
DPOBackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Overlay, Visual Prompting=Default2025.079.660.851.2
NDBackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Inductive, Visual Prompting=Default2025.079.56555.5
DPOBackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Inductive, Visual Prompting=Highlight2025.076.568.562
DPOBackbone=Qwen2-vl-7B (LaSM L7-18, α=1.1), Injection Type (IT)=Inductive, Visual Prompting=Default2025.07662.556.5