AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Mind the Gap: Quantifying and Aligning Human-AI Visual Attention for Accident Anticipation

Hoe Sung Ryu, Christian Wallraven

AAAI 2026 · Vision-language and multimodal › Human and embodied modeling · approach: Human-like visual cognition in VLMs · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add prompt-based attention guidance (ablation)VLMs · Attention-DADAmacro F1: open-source about +7%, closed-source +6.5%Improved
add visual-based human gaze guidance (ablation)VLMs · Attention-DADAmacro F1: open-source about +4%, closed-source <1%Improved
add prompt guidance to Gemini-2.5-Flash (ablation)Gemini-2.5-Flash · Attention-DADA T1macro F1: 0.62 -> 0.67Improved
add prompt guidance to QwenVL-2.5 (ablation)QwenVL-2.5 · Attention-DADA T1macro F1: 0.42 -> 0.43No clear effect
add attention guidance (prompt or visual) (ablation)VLMs · Attention-DADAgap to human F1 near event: human 0.88 vs best VLM 0.67 at T1No clear effect
add attention guidance (prompt or visual) (ablation)InternVL-3 · Attention-DADAKendall's tau alignmentNo clear effect
move from earliest (T5) to latest (T1) cliphuman participants · Attention-DADAmacro F1: 0.46 -> 0.88Improved
move from earliest (T5) to latest (T1) clipzero-shot VLMs · Attention-DADAmacro F1: e.g. QwenVL-2.5 0.38 -> 0.42No clear effect
accuracy level (high vs low) for prompt-guided model on normal videosprompt-guided VLM · Attention-DADA normal videosKendall's tau: 0.355 -> 0.481Improved
accuracy level vs tau on other five conditionsVLMs · Attention-DADAKendall's tauNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy