Research map / Graph learning / Graph self-supervised learning
Regularized Primitive Graph Learning for Unified Vector Mapping
Lei Wang, Min Dai, Jianan He, Jingwei Huang
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 shape regularization loss Lreg (ablation) | GraphMapper (building) · CrowdAI | C-IoU: 82.7 -> 89.5 | Improved |
| add direction regression loss Ldir (ablation) | GraphMapper (building) · CrowdAI | MTA: 33.1° -> 30.4° | Improved |
| add auxiliary loss Laux (ablation) | GraphMapper (building) · CrowdAI | C-IoU: 89.0 -> 89.5 | Improved |
| use incremental (sequential) refinement then relationship reconstruction instead of parallel (ablation) | GraphMapper (building) · CrowdAI | C-IoU: 78.3 -> 89.5 | Improved |
| use incremental reconstruction (ablation) | GraphMapper (road) · City-Scale | APLS: 67.6 -> 68.9 | Improved |
| sort points in embedding space for connections instead of thresholding probability (ablation) | GraphMapper (road) · City-Scale | APLS: 66.2 -> 68.9 | Improved |
| remove Lreg on IoU (ablation) | GraphMapper (building) · CrowdAI | IoU: 94.4 vs 94.3 | No clear effect |
| replace PolyWorld with GraphMapper | building mapping · CrowdAI | AP: 63.3 -> 72.8 | Improved |
| replace Sat2Graph with GraphMapper | road mapping · City-Scale | TOPO F1: 76.3 -> 85.6 | Improved |
| add primitive refinement and relationship reconstruction to segmentation mask (ablation) | GraphMapper (building) · CrowdAI | AP: 63.6 -> 72.8 | 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.
- Sat2Graph — Sat2Graph: Road Graph Extraction through Graph-Tensor Encoding (ECCV 2020) · better on 4 of 4 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
- Topological Map Extraction From Overhead Images (ICCV 2019 · similarity 0.88)
- Patched Line Segment Learning for Vector Road Mapping (AAAI 2024 · similarity 0.87)
- Vectorizing World Buildings: Planar Graph Reconstruction by Primitive Detection and Relationship Inference (ECCV 2020 · similarity 0.86)
- LineGraph2Road: Structural Graph Reasoning on Line Graphs for Road Network Extraction (ECCV 2026 · similarity 0.86)
- Sat2Graph: Road Graph Extraction through Graph-Tensor Encoding (ECCV 2020 · similarity 0.86)
- VectorMapNet: End-to-end Vectorized HD Map Learning (ICML 2023 · similarity 0.85)
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.
