Research map / Detection and segmentation / CNN features
Learning Global Representation from Queries for Vectorized HD Map Construction
Shoumeng Qiu, Xinrun Li, Yang Long, Xiangyang Xue, Varun Ojha, Jian Pu
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 GRL and GRG plug-in to MapQR | MapQR · nuScenes val 24 epochs | mAP2: 65.3 -> 68.1 | Improved |
| add GRL and GRG plug-in to MapQR at 110 epochs | MapQR · nuScenes val 110 epochs | mAP2: 72.6 -> 73.1 | Improved |
| add GRL and GRG plug-in to MapTRv2 | MapTRv2 · nuScenes val 24 epochs | mAP2: 61.5 -> 65.0 | Improved |
| add GRL only (without GRG) (ablation) | MapTRv2 · nuScenes val 24 epochs | mAP2: 61.5 -> 63.7 | Improved |
| add GRG on top of GRL (ablation) | MapTRv2 + GRL · nuScenes val 24 epochs | mAP2: 63.7 -> 65.0 | Improved |
| add MapGR to MapQR on Argoverse 2 | MapQR · Argoverse 2 | mAP2: 65.1 -> 66.0 | Improved |
| GRL MLP dimension 2048 vs 512 or 8192 (ablation) | MapGR · nuScenes val | mAP2: 64.6 / 65.0 / 64.2 | Mixed |
| apply GRL+GRG to first 2 layers instead of 1 or 3 (ablation) | MapQR + MapGR · nuScenes val | mAP2: 67.03 / 68.12 / 67.11 for N=1/2/3 | Mixed |
| vary GRL loss weight over an order of magnitude (ablation) | MapQR + MapGR · nuScenes val | mAP: 67.7 / 68.1 / 67.6 | No clear effect |
| add GRG alone without new loss (ablation) | MapQR · nuScenes val | mAP: 65.3 -> 66.9 | Improved |
| replace BEV-feature seg head with query-driven GRL head (ablation) | MapQR · nuScenes val | mAP: 65.3 -> 67.5 | Improved |
| re-enable GRL+GRG at layer 2 with 4x mask resolution (ablation) | MapQR + MapGR · nuScenes val | mAP2: 68.1 -> 68.5 | Improved |
Closest papers by meaning
- InteractionMap: Improving Online Vectorized HDMap Construction with Interaction (CVPR 2025 · similarity 0.88)
- Uncertainty-Instructed Structure Injection for Generalizable HD Map Construction (CVPR 2025 · similarity 0.88)
- MapExpert: Online HD Map Construction with Simple and Efficient Sparse Map Element Expert (AAAI 2025 · similarity 0.87)
- SDTagNet: Leveraging Text-Annotated Navigation Maps for Online HD Map Construction (NeurIPS 2025 · similarity 0.87)
- Unveiling the Hidden: Online Vectorized HD Map Construction with Clip-Level Token Interaction and Propagation (NeurIPS 2024 · similarity 0.86)
- HIMap: HybrId Representation Learning for End-to-end Vectorized HD Map Construction (CVPR 2024 · similarity 0.86)
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.
