Research map / Video / Video-language understanding
Traffic Accident Benchmark for Causality Recognition
Tackgeun You, Bohyung Han
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 |
|---|---|---|---|
| use linear consensus instead of average consensus (ablation) | TSN · CTA classification (mean, top-1) | Top-1 mean accuracy (%): 31.3 -> 59.4 | Improved |
| use average consensus (vs trivial prediction) (ablation) | Trivial Prediction · CTA classification (mean, top-1) | Top-1 mean accuracy (%): 46.9 -> 31.3 | Hurt |
| replace forward GRU with backward GRU (ablation) | SST · CTA localization (mean, tIoU>0.3) | Accuracy (%): 27.59 -> 37.41 | Improved |
| replace forward GRU with bi-directional GRU (ablation) | SST · CTA localization (mean, tIoU>0.3) | Accuracy (%): 27.59 -> 42.07 | Improved |
| replace GRU with SS-TCN dilated convolutions (ablation) | SST · CTA localization (mean, tIoU>0.7) | Accuracy (%): 7.59 -> 11.03 | Improved |
| replace GRU with SS-TCN dilated convolutions (ablation) | Bi-SST · CTA localization (mean, tIoU>0.3) | Accuracy (%): 42.07 -> 40.69 | Improved |
| stack multiple stages (MS-TCN vs SS-TCN) (ablation) | SS-TCN · CTA localization (mean, tIoU>0.5) | Accuracy (%): 30.17 -> 32.97 | Improved |
| use segmentation (MS-TCN) instead of detection (R-C3D) (ablation) | R-C3D · CTA localization (mean, tIoU>0.7) | Accuracy (%): 10.86 -> 14.41 | Improved |
| normalize video lengths for trivial prediction (ablation) | Trivial prediction · CTA localization (mean, tIoU>0.3) | Accuracy (%): 20.00 -> 29.48 | Improved |
Closest papers by meaning
- Accident Anticipation via Temporal Occurrence Prediction (NeurIPS 2025 · similarity 0.90)
- Abductive Ego-View Accident Video Understanding for Safe Driving Perception (CVPR 2024 · similarity 0.86)
- Progressive Risk Estimation for Accident Anticipation (NeurIPS 2026 · similarity 0.85)
- TraffiDent: A Dataset for Understanding the Interplay Between Traffic Dynamics and Incidents (NeurIPS 2025 · similarity 0.85)
- MECD: Unlocking Multi-Event Causal Discovery in Video Reasoning (NeurIPS 2024 · similarity 0.83)
- TUMTraf VideoQA: Dataset and Benchmark for Unified Spatio-Temporal Video Understanding in Traffic Scenes (ICML 2025 · similarity 0.83)
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.
