Research map / Probabilistic and scientific modeling / Molecules and proteins
From Gaze to Movement: Predicting Visual Attention for Autonomous Driving Human-Machine Interaction based on Programmatic Imitation Learning
Yexin Huang, Yongbin Lin, Lishengsa Yue, Zhihong Yao, Jie Wang
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 |
|---|---|---|---|
| selectively merge connected vehicle segments into one gaze target (ablation) | PILOT · DATAD (Subject 1) | MSE per frame: 2,511,482 -> 634,401 | Improved |
| selectively merge connected vehicle segments into one gaze target (ablation) | PILOT · DATAD (Subject 1) | Average hit rate: 72.13% -> 88.72% | Improved |
| selectively merge connected vehicle segments into one gaze target (ablation) | PILOT · DATAD (Subject 1) | At-least-once hit rate: 95.82% vs 95.50% | No clear effect |
| add left obj distance feature (ablation) | PILOT · DATAD (Subject 1) | MSE per frame: 2,915,135 -> 634,401 | Improved |
| add left obj distance feature (ablation) | PILOT · DATAD (Subject 1) | Average hit rate: 68.75% -> 88.72% | Improved |
| add brake feature (ablation) | PILOT · DATAD (Subject 1) | MSE per frame: 1,972,310 -> 634,401 | Improved |
| add fix obj speed feature (ablation) | PILOT · DATAD (Subject 1) | MSE per frame: 718,895 -> 634,401 | Improved |
| add right obj speed feature (ablation) | PILOT · DATAD (Subject 1) | Average hit rate: 82.55% -> 88.72% | Improved |
| add ego car speed feature (ablation) | PILOT · DATAD (Subject 1) | At-least-once hit rate: 91.00% -> 95.50% | Improved |
| add right obj distance feature (ablation) | PILOT · DATAD (Subject 1) | At-least-once hit rate: 91.67% -> 95.50% | Improved |
| use PILOT instead of neural saliency baselines | PILOT vs SalM2 · DATAD | MSE per frame: 1,033,065 -> 634,401 | Improved |
| use PILOT instead of PGNet | PILOT vs PGNet · DATAD | At-least-once hit rate: 96.08% vs 95.50% | No clear effect |
Closest papers by meaning
- Leveraging Driver Field-of-View for Multimodal Ego-Trajectory Prediction (ICLR 2025 · similarity 0.88)
- FIFA: Fine-grained Inter-frame Attention for Driver's Video Gaze Estimation (CVPR 2025 · similarity 0.87)
- Continuously Learning, Adapting, and Improving: A Dual-Process Approach to Autonomous Driving (NeurIPS 2024 · similarity 0.87)
- ST-DiffEye: Diffusion-based Continuous Gaze Generation via Joint Scanpath-Trajectory Modeling (NeurIPS 2026 · similarity 0.87)
- BAT: Behavior-Aware Human-Like Trajectory Prediction for Autonomous Driving (AAAI 2024 · similarity 0.87)
- EyEar: Learning Audio Synchronized Human Gaze Trajectory Based on Physics-Informed Dynamics (AAAI 2025 · 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.
