Research map / Multi-agent systems and games / Human-AI and goal-conditioned
DRIVE: Deep Reinforced Accident Anticipation With Visual Explanation
Wentao Bao, Qi Yu, Yu Kong
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 fixation prediction policy and fixation reward (ablation) | DRIVE · DADA-2000 | AUC (%): 61.91 -> 72.27 | Improved |
| add regularized auto-encoder (RAE) for the state (ablation) | DRIVE · DADA-2000 | AUC (%): 66.21 -> 72.27 | Improved |
| train with SAC reinforcement learning instead of supervised learning (ablation) | DRIVE · DADA-2000 | AUC (%): 63.96 -> 72.27 | Improved |
| use dynamic attention fusion instead of static (param 0.5) (ablation) | DRIVE · DADA-2000 | AUC: 0.645 -> 0.659 | Improved |
| use dynamic attention fusion instead of static (param 0.8) (ablation) | DRIVE · DADA-2000 | AUC: 0.691 -> 0.726 | Improved |
| use dynamic attention fusion instead of static (param 1.0) (ablation) | DRIVE · DADA-2000 | AUC: 0.632 -> 0.679 | Improved |
| use dynamic attention fusion instead of static (param 1.0) (ablation) | DRIVE · DADA-2000 | KLD: 12.948 -> 7.836 | Improved |
| use DRIVE instead of AdaLEA on earliness | AdaLEA · DADA-2000 | TTA (s): 3.890 -> 3.657 | Hurt |
| use DRIVE instead of UString | UString · DAD | AUC (%): 65.96 -> 93.82 | Improved |
| use DRIVE instead of DSA-RNN | DSA-RNN · DAD | TTA (s): 1.169 -> 2.781 | Improved |
| remove attention at test time (S=1) or invert it (ablation) | DRIVE · DADA-2000 | recall and f-AUC | Hurt |
Closest papers by meaning
- Progressive Risk Estimation for Accident Anticipation (NeurIPS 2026 · similarity 0.87)
- Towards Safer and Understandable Driver Intention Prediction (ICCV 2025 · similarity 0.87)
- Accident Anticipation via Temporal Occurrence Prediction (NeurIPS 2025 · similarity 0.84)
- Abductive Ego-View Accident Video Understanding for Safe Driving Perception (CVPR 2024 · similarity 0.84)
- Driving into the Future: Multiview Visual Forecasting and Planning with World Model for Autonomous Driving (CVPR 2024 · similarity 0.84)
- SafeAuto: Knowledge-Enhanced Safe Autonomous Driving with Multimodal Foundation Models (ICML 2025 · similarity 0.84)
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.
