Research map / Graph learning / Knowledge and temporal graphs
Visual Traffic Knowledge Graph Generation from Scene Images
Yunfei Guo, Fei Yin, Xiao-hui Li, Xudong Yan, Tao Xue, Shuqi Mei, Cheng-Lin Liu
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 hierarchical three-level graph instead of a homogeneous graph with only BiAtt (ablation) | HGAT · RS10K test | Mean relation F1: 0.807 -> 0.841 | Improved |
| add cross-level links and CLAtt (ablation) | HGAT · RS10K test | Mean relation F1: 0.831 -> 0.841 | Improved |
| use predicted S-S relations to filter C-C relations rather than heuristic rules (ablation) | HGAT · RS10K test | C-C relation F1: 0.800 -> 0.821 | Improved |
| add T-T links between traffic elements (ablation) | HGAT · RS10K test | Mean relation F1: 0.827 -> 0.841 | Improved |
| add semantic node feature (ablation) | HGAT · RS10K test | Mean relation F1: 0.820 -> 0.841 | Improved |
| add spatial node feature (ablation) | HGAT · RS10K test | Mean relation F1: 0.829 -> 0.841 | Improved |
| replace GAT/GCN/HL-Net message passing with HGAT | relation reasoning · RS10K test | Mean relation F1: 0.825 -> 0.841 | Improved |
| apply RoIAlign window features with rotated FCOS head for sign components | sign component detector · RS10K test | Detection F1: 0.853 -> 0.903 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- HL-Net — HL-Net: Heterophily Learning Network for Scene Graph Generation (CVPR 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- QDNet: Query-Denoising Network for Visual Traffic Knowledge Graph Generation (TPAMI 2025 · similarity 0.91)
- Traffic Scene Parsing through the TSP6K Dataset (CVPR 2024 · similarity 0.89)
- T2SG: Traffic Topology Scene Graph for Topology Reasoning in Autonomous Driving (CVPR 2025 · similarity 0.86)
- TopoLogic: An Interpretable Pipeline for Lane Topology Reasoning on Driving Scenes (NeurIPS 2024 · similarity 0.85)
- Efficient Traffic Prediction Through Spatio-Temporal Distillation (AAAI 2025 · similarity 0.84)
- RoadSceneBench: A Lightweight Benchmark for Mid-Level Road Scene Understanding (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.
