Research map / Vision-language and multimodal / Robot action and navigation
From Gaze to Meaning: An AI Agent for Unified Zero-Shot Grounding and Explanation
Shayan Nasiriboukani, Sara Atito, Mohammad Nezamipour, Muhammad Awais
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add uncertainty-gated memory retrieval to GGVL (ablation) | GGVL · GazeFollow | Acc@1: 0.450 -> 0.493 | Improved |
| add uncertainty-gated retrieval (ablation) | GGVL · GazeHOI | Acc@3: 0.721 -> 0.745 | Improved |
| add retrieval to LLaVA 1.5 4B (ablation) | GGVL · GazeFollow | Acc@1: 0.380 -> 0.466 | Improved |
| add retrieval to Gemini 2.5 Pro (ablation) | GGVL · GazeFollow | Acc@1: 0.584 -> 0.608 | Improved |
| add retrieval to Qwen3-VL 8B (ablation) | GGVL · GazeFollow | Acc@1: 0.584 -> 0.602 | Improved |
| retrieve for all samples instead of dynamic uncertainty threshold (ablation) | GTA · GazeFollow (ablation) | Acc@1: 48.80 -> 42.64 | Hurt |
| fixed threshold 25 instead of dynamic threshold (ablation) | GTA · GazeFollow (ablation) | Acc@1: 48.80 -> 48.49 | No clear effect |
| use gaze arrow instead of head bounding box prompt (ablation) | GTA · GazeFollow (ablation) | Acc@1: 28.69 -> 48.80 | Improved |
| use arrow instead of blue dot prompt (ablation) | GTA · GazeFollow (ablation) | Acc@1: 40.99 -> 48.80 | Improved |
| widen blue arrow from 4px to 5px (ablation) | GTA · GazeFollow (ablation) | Acc@1: 48.80 -> 49.29 | Improved |
| use k=6 retrieved examples instead of k=3 (ablation) | GTA · GazeFollow (ablation) | Acc@1: 48.68 -> 49.29 | Improved |
| use k=6 instead of k=10 retrieved examples (ablation) | GTA · GazeFollow (ablation) | Acc@1: 49.04 -> 49.29 | Improved |
Closest papers by meaning
- Eyes on VLM: Benchmarking Gaze Following and Social Gaze Prediction in Vision Language Models (NeurIPS 2026 · similarity 0.87)
- OmniGF: A Dual-Branch Vision-Language Framework for Unified Gaze Following (NeurIPS 2026 · similarity 0.87)
- Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning (AAAI 2026 · similarity 0.87)
- Why Is Spatial Reasoning Hard for VLMs? An Attention Mechanism Perspective on Focus Areas (ICML 2025 · similarity 0.86)
- DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models (CVPR 2026 · similarity 0.86)
- DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning (ICLR 2026 · similarity 0.86)
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.
