Research map / Vision-language and multimodal / Human and embodied modeling
GazeXplain: Learning to Predict Natural Language Explanations of Visual Scanpaths
Xianyu Chen, Ming Jiang, Qi Zhao
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 semantic alignment loss (ablation) | GazeXplain · AiR-D | CIDEr-R: 97.3 -> 123.1 | Improved |
| add semantic alignment loss (ablation) | GazeXplain · AiR-D | ScanMatch (SM): 0.378 -> 0.386 | Improved |
| add the language decoder for explanations (with co-training) (ablation) | GazeXplain · AiR-D | ScanMatch (SM): 0.356 -> 0.378 | Improved |
| add the language decoder for explanations (no co-training, no alignment) (ablation) | GazeXplain · AiR-D | ScanMatch (SM): 0.337 -> 0.339 | No clear effect |
| add the language decoder for explanations (ablation) | GazeXplain · AiR-D | CIDEr-R: 66.7 -> 97.3 | Improved |
| add cross-dataset co-training (ablation) | GazeXplain · AiR-D | ScanMatch (SM): 0.346 -> 0.386 | Improved |
| add semantic alignment loss (ablation) | GazeXplain · AiR-D | BLEU-4: 27.7 -> 30.7 | Improved |
| add semantic alignment loss (ablation) | GazeXplain · COCO-Search18 target-absent | vocabulary size: 566 -> 685 | Improved |
| add semantic alignment loss (ablation) | GazeXplain · COCO-Search18 explanation faithfulness | ROC AUC: 0.878 -> 0.902 | Improved |
| add cross-dataset co-training (ablation) | GazeXplain · COCO-Search18 explanation faithfulness | ROC AUC: 0.871 -> 0.902 | Improved |
| use GazeXplain versus ChenLSTM on target-present search | GazeXplain · COCO-Search18 target-present | SED: ChenLSTM 1.932 vs GazeXplain 1.981 (SED); SM 0.448 vs 0.480 | Mixed |
Later papers that compared with this method
1 later accepted paper reports GazeXplain as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Explainable Saliency: Articulating Reasoning with Contextual Prioritization (CVPR 2025 · better on 3 of 4 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Eyes on VLM: Benchmarking Gaze Following and Social Gaze Prediction in Vision Language Models (NeurIPS 2026 · similarity 0.89)
- Unifying Top-down and Bottom-up Scanpath Prediction Using Transformers (CVPR 2024 · similarity 0.88)
- EyEar: Learning Audio Synchronized Human Gaze Trajectory Based on Physics-Informed Dynamics (AAAI 2025 · similarity 0.88)
- OmniGF: A Dual-Branch Vision-Language Framework for Unified Gaze Following (NeurIPS 2026 · similarity 0.88)
- ST-DiffEye: Diffusion-based Continuous Gaze Generation via Joint Scanpath-Trajectory Modeling (NeurIPS 2026 · similarity 0.87)
- Pathways of Visual Information Flow in Vision-Language Models (NeurIPS 2026 · similarity 0.87)
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.
