Research map / Detection and segmentation / Open-set and few-shot detection
Connecting Gaze, Scene, and Attention: Generalized Attention Estimation via Joint Modeling of Gaze and Scene Saliency
Eunji Chong, Nataniel Ruiz, Yongxin Wang, Yun Zhang, Agata Rozga, James M. Rehg
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 EYEDIAP training data (ablation) | gaze-saliency model · GazeFollow test | AUC: 0.887 -> 0.896 | Improved |
| add SynHead training data (ablation) | gaze-saliency model · GazeFollow test | AUC: 0.895 -> 0.896 | No clear effect |
| add both EYEDIAP and SynHead (ablation) | gaze-saliency model · GazeFollow test | AUC: 0.891 -> 0.896 | Improved |
| add project-and-compare loss (ablation) | gaze-saliency model · GazeFollow test | AUC: 0.895 -> 0.896 | No clear effect |
| use 10x10 heatmap instead of 15x15 map resolution (ablation) | gaze-saliency model · GazeFollow test | AUC: 0.778 -> 0.896 | Improved |
| use separate scene pathway plus face pathway instead of single ResNet50 with ROI-pooling (ablation) | gaze-saliency model · GazeFollow test | AUC: 0.700 -> 0.896 | Improved |
| add EYEDIAP to GazeFollow-only training (ablation) | fixation likelihood model · MMDB fixation likelihood | average precision: 0.737 -> 0.820 | Improved |
| add SynHead to GazeFollow and EYEDIAP training (ablation) | fixation likelihood model · MMDB fixation likelihood | average precision: 0.820 -> 0.902 | Improved |
| use joint gaze and saliency modeling vs GazeFollow-only baseline on 2x2 grid (ablation) | gaze-saliency model · MMDB grid classification 2x2 | precision: 0.344 -> 0.744 | Improved |
| use joint gaze and saliency modeling vs GazeFollow baseline on 5x5 grid (ablation) | gaze-saliency model · MMDB grid classification 5x5 | precision: 0.210 -> 0.614 | Improved |
| add gaze angle features to SVM with GazeFollow (ablation) | SVM classifier · MMDB fixation likelihood | average precision: 0.311 -> 0.531 | Improved |
| add head pose features to Random Forest with GazeFollow (ablation) | Random Forest classifier · MMDB fixation likelihood | average precision: 0.707 -> 0.785 | Improved |
Closest papers by meaning
- Learning from Observer Gaze: Zero-Shot Attention Prediction Oriented by Human-Object Interaction Recognition (CVPR 2024 · similarity 0.90)
- Learning From Unique Perspectives: User-Aware Saliency Modeling (CVPR 2023 · similarity 0.90)
- Gazeformer: Scalable, Effective and Fast Prediction of Goal-Directed Human Attention (CVPR 2023 · similarity 0.90)
- MTGS: A Novel Framework for Multi-Person Temporal Gaze Following and Social Gaze Prediction (NeurIPS 2024 · similarity 0.89)
- CLIP-Gaze: Towards General Gaze Estimation via Visual-Linguistic Model (AAAI 2024 · similarity 0.88)
- Eyes on VLM: Benchmarking Gaze Following and Social Gaze Prediction in Vision Language Models (NeurIPS 2026 · similarity 0.88)
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.
