Research map / Graph learning / Knowledge and temporal graphs
Lane Graph as Path: Continuity-preserving Path-wise Modeling for Online Lane Graph Construction
Bencheng Liao, Shaoyu Chen, Bo Jiang, Tianheng Cheng, Qian Zhang, Wenyu Liu, Chang Huang, Xinggang Wang
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 |
|---|---|---|---|
| model the lane graph as continuous paths instead of pieces (MapTR with same decoder) | MapTR piece-wise · nuScenes val | Junction TOPO F1: 0.496 -> 0.564 | Improved |
| path-wise modeling versus pixel-wise segmentation modeling | HDMapNet · nuScenes val | TOPO F1 and FPS: TOPO F1 0.513 -> 0.529; FPS 6.8 -> 15.6 | Improved |
| adapt TopoNet lane detection to path-wise representation | TopoNet · OpenLane-V2 | mIoU: 39.0 -> 40.6 | Improved |
| add auxiliary BEV segmentation loss (ablation) | LaneGAP · nuScenes val (24 epochs) | Junction TOPO F1: 0.529 -> 0.539 | Improved |
| increase polyline points from 10 to 40 (ablation) | LaneGAP · nuScenes val (24 epochs) | Junction TOPO F1: 0.532 -> 0.557 | Improved |
| increase Bezier control points from 5 to 10 (ablation) | LaneGAP · nuScenes val (24 epochs) | Junction TOPO F1: 0.510 -> 0.501 | Hurt |
| increase Bezier control points from 3 to 5 (ablation) | LaneGAP · nuScenes val (24 epochs) | Junction TOPO F1: 0.481 -> 0.510 | Improved |
| replace MapTR centerline map branch with LaneGAP path queries in end-to-end planner | VAD-Tiny · nuScenes planning | avg L2 (m): 0.74 -> 0.70 | Improved |
| add LiDAR modality to vision (ablation) | LaneGAP · nuScenes (24 epochs) | Junction TOPO F1: 0.539 -> 0.615 | Improved |
| train longer (110 vs 24 epochs) (ablation) | LaneGAP vision-only · nuScenes | Junction TOPO F1 vs TOPO F1: Junction 0.539 -> 0.564; TOPO 0.461 -> 0.529 | Mixed |
| use path-wise instead of pixel-wise modeling on precision of junction TOPO | HDMapNet · Argoverse2 | Junction TOPO precision: 0.656 -> 0.663 | No clear effect |
Closest papers by meaning
- LaneSegNet: Map Learning with Lane Segment Perception for Autonomous Driving (ICLR 2024 · similarity 0.90)
- LaneGraph2Seq: Lane Topology Extraction with Language Model via Vertex-Edge Encoding and Connectivity Enhancement (AAAI 2024 · similarity 0.90)
- LaneCPP: Continuous 3D Lane Detection using Physical Priors (CVPR 2024 · similarity 0.89)
- GLane3D: Detecting Lanes with Graph of 3D Keypoints (CVPR 2025 · similarity 0.89)
- T2SG: Traffic Topology Scene Graph for Topology Reasoning in Autonomous Driving (CVPR 2025 · similarity 0.88)
- Bezier Everywhere All at Once: Learning Drivable Lanes as Bezier Graphs (CVPR 2024 · similarity 0.88)
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.
