Research map / Detection and segmentation / CNN features
Explainable Object-Induced Action Decision for Autonomous Vehicles
Yiran Xu, Xiaoyin Yang, Lihang Gong, Hsuan-Chu Lin, Tz-Ying Wu, Yunsheng Li, Nuno Vasconcelos
Check 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 explanation supervision with equal weight to action loss (lambda 1.0), rather than train actions alone (ablation) | Proposed network · BDD-OIA test | action F1_all: 0.675 -> 0.734 | Improved |
| Increase explanation loss weight from 0 to 0.01 (ablation) | Proposed network · BDD-OIA test | action F1_all: 0.675 -> 0.696 | Improved |
| Increase explanation loss weight from 0.01 to 0.1 (ablation) | Proposed network · BDD-OIA test | action F1_all: 0.696 -> 0.709 | Improved |
| Increase explanation loss weight from 0.1 to 1.0 (ablation) | Proposed network · BDD-OIA test | action F1_all: 0.709 -> 0.734 | Improved |
| Remove global scene features (ablation) | Proposed network · BDD-OIA test | action F1_all: 0.734 -> 0.605 | Hurt |
| Remove local object features (ablation) | Proposed network · BDD-OIA test | action F1_all: 0.734 -> 0.704 | Hurt |
| Replace learned object selection with random selection (ablation) | Proposed network · BDD-OIA test | action F1_all: 0.734 -> 0.709 | Hurt |
| Select top 5 rather than top 10 action-inducing objects (ablation) | Proposed network · BDD-OIA test | action F1_all: 0.720 vs. 0.734 | Hurt |
| Use one action label instead of multiple possible-action labels (ablation) | Proposed network · BDD-OIA test | action mF1: 0.648 -> 0.455 | Hurt |
| Use one action label instead of multiple possible-action labels (ablation) | Proposed network · BDD-OIA test | action F1_all: 0.665 -> 0.715 | Improved |
| Combine local object features with global scene context (ablation) | Proposed network · BDD-OIA test | action F1_all: 0.704 -> 0.734 | Improved |
| Add local object features to the global branch (ablation) | Proposed network · BDD-OIA test | explanation F1_all: 0.419 -> 0.422 | No clear effect |
Closest papers by meaning
- InAction: Interpretable Action Decision Making for Autonomous Driving (ECCV 2022 · similarity 0.88)
- Towards Safer and Understandable Driver Intention Prediction (ICCV 2025 · similarity 0.88)
- Textual Explanations for Self-Driving Vehicles (ECCV 2018 · similarity 0.88)
- DualAD: Disentangling the Dynamic and Static World for End-to-End Driving (CVPR 2024 · similarity 0.84)
- Continuously Learning, Adapting, and Improving: A Dual-Process Approach to Autonomous Driving (NeurIPS 2024 · similarity 0.84)
- SGDrive: Scene-to-Goal Hierarchical World Cognition for Autonomous Driving (CVPR 2026 · 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.
