Research map / Graph learning / Knowledge and temporal graphs
Continuity Preserving Online CenterLine Graph Learning
Yunhui Han, Kun Yu, Zhiwei Li
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 Junction Aware Query Enhancement (JAQ) (ablation) | modified MapTR baseline · nuScenes val (24 epochs) | APLS: 26.8 -> 28.5 | Improved |
| add Bezier Space Connection (BSC) (ablation) | modified MapTR baseline · nuScenes val (24 epochs) | APLS: 26.8 -> 29.1 | Improved |
| add Iterative Topology Refinement (ITR) (ablation) | modified MapTR baseline · nuScenes val (24 epochs) | APLS: 26.8 -> 28.6 | Improved |
| combine JAQ, BSC and ITR (ablation) | modified MapTR baseline · nuScenes val (24 epochs) | TOPO F1: 39.7 -> 42.2 | Improved |
| increase number of Bezier control points (ablation) | CGNet · nuScenes val (24 epochs) | TOPO F1: 42.2 -> 40.4 (4 to 8 points) | Hurt |
| use junction dilation radius 9 vs 7 or 11 (ablation) | CGNet · nuScenes val (24 epochs) | TOPO F1: 40.4 (R=7) -> 42.2 (R=9) | Improved |
| project lane embeddings (feature) instead of final polyline (lane) into Bezier space (ablation) | CGNet · nuScenes val (24 epochs) | TOPO F1: 40.1 -> 42.2 | Improved |
| increase ITR iterations from 1 to 6 (ablation) | CGNet · nuScenes val (24 epochs) | TOPO F1: 39.6 -> 42.2 | Improved |
| replace MapTR with CGNet at 110 epochs | MapTR · nuScenes val (110 epochs) | APLS: 38.4 -> 41.4 | Improved |
| replace MapTR with CGNet on Argoverse2 | MapTR · Argoverse2 val | SDA: 13.6 -> 13.7 | No clear effect |
Closest papers by meaning
- LaneGraph2Seq: Lane Topology Extraction with Language Model via Vertex-Edge Encoding and Connectivity Enhancement (AAAI 2024 · similarity 0.86)
- TopoMLP: A Simple yet Strong Pipeline for Driving Topology Reasoning (ICLR 2024 · similarity 0.85)
- Topo2Seq: Enhanced Topology Reasoning via Topology Sequence Learning (AAAI 2025 · similarity 0.85)
- Bezier Everywhere All at Once: Learning Drivable Lanes as Bezier Graphs (CVPR 2024 · similarity 0.84)
- TopoPoint: Enhance Topology Reasoning via Endpoint Detection in Autonomous Driving (NeurIPS 2025 · similarity 0.84)
- TopoCurve: Geometry-Aware Topology Reasoning via Bézier Curves in Autonomous Driving (NeurIPS 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.
