Research map / Point clouds / Point cloud classification
TopoHR: Hierarchical Centerline Representation for Cyclic Topology Reasoning in Driving Scenes with Point-to-Instance Relations
Yifeng Bai, Zhirong Chen, Bo Song, Erkang Cheng, Haibin Ling
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 point-level queries to instance queries (ablation) | TopoHR representation · OpenLane-V2 subset_A (ablation) | DET_l: 26.8 -> 19.6 | Hurt |
| add P2P constrained relation mask (ablation) | Ins+Pts · OpenLane-V2 subset_A (ablation) | DET_l: 19.6 -> 20.1 | Improved |
| add hierarchical integrator (ablation) | Ins+Pts+P2P · OpenLane-V2 subset_A (ablation) | DET_l: 20.1 -> 32.2 | Improved |
| add discrete distance transform segmentation instead of binary segmentation GT (ablation) | TopoHR · OpenLane-V2 subset_A (ablation) | DET_l: 32.6 -> 34.6 | Improved |
| use distance transform instead of binary mask (ablation) | TopoHR · OpenLane-V2 subset_A (ablation) | DET_l: 32.6 -> 32.0 | No clear effect |
| add I2I cyclic forward and backward feedback (ablation) | TopoHR instance queries · OpenLane-V2 subset_A (ablation) | DET_l: 34.8 -> 35.6 | Improved |
| add point queries and P2I relations to the cyclic pipeline (ablation) | TopoHR · OpenLane-V2 subset_A (ablation) | TOP_lt: 32.9 -> 34.6 | Improved |
| replace focal loss with adaptive topological loss (ablation) | TopoHR · OpenLane-V2 subset_A (ablation) | DET_l: 34.9 -> 36.1 | Improved |
| use dice loss instead of focal loss for topology (ablation) | TopoHR · OpenLane-V2 subset_A (ablation) | DET_l: 34.9 -> 30.4 | Hurt |
| raise lambda_pos from 200 to 400 in ATL (ablation) | TopoHR · OpenLane-V2 subset_A (ablation) | DET_l: 33.8 -> 36.1 | Improved |
| raise lambda_neg from 5 to 10 in ATL (ablation) | TopoHR · OpenLane-V2 subset_A (ablation) | DET_l: 36.1 -> 34.0 | Hurt |
| extend training from 24 to 48 epochs (ablation) | TopoHR-L · OpenLane-V2 subset_B | TOP_ll: 35.8 -> 39.7 | 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.
- TopoPoint — TopoPoint: Enhance Topology Reasoning via Endpoint Detection in Autonomous Driving (NeurIPS 2025) · better on 3 of 3 shared results
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
- Topo2Seq: Enhanced Topology Reasoning via Topology Sequence Learning (AAAI 2025 · similarity 0.93)
- TopoMLP: A Simple yet Strong Pipeline for Driving Topology Reasoning (ICLR 2024 · similarity 0.92)
- T2SG: Traffic Topology Scene Graph for Topology Reasoning in Autonomous Driving (CVPR 2025 · similarity 0.91)
- TopoPoint: Enhance Topology Reasoning via Endpoint Detection in Autonomous Driving (NeurIPS 2025 · similarity 0.91)
- TopoLogic: An Interpretable Pipeline for Lane Topology Reasoning on Driving Scenes (NeurIPS 2024 · similarity 0.91)
- TopoCurve: Geometry-Aware Topology Reasoning via Bézier Curves in Autonomous Driving (NeurIPS 2026 · similarity 0.90)
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.
