Reusing Attention for One-stage Lane Topology Understanding
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909702408896512 |
|---|---|
| author | Li, Yang Zhang, Zongzheng Qiu, Xuchong Li, Xinrun Liu, Ziming Wang, Leichen Li, Ruikai Zhu, Zhenxin Gao, Huan-ang Lin, Xiaojian Cui, Zhiyong Zhao, Hang Zhao, Hao |
| author_facet | Li, Yang Zhang, Zongzheng Qiu, Xuchong Li, Xinrun Liu, Ziming Wang, Leichen Li, Ruikai Zhu, Zhenxin Gao, Huan-ang Lin, Xiaojian Cui, Zhiyong Zhao, Hang Zhao, Hao |
| contents | Understanding lane toplogy relationships accurately is critical for safe autonomous driving. However, existing two-stage methods suffer from inefficiencies due to error propagations and increased computational overheads. To address these challenges, we propose a one-stage architecture that simultaneously predicts traffic elements, lane centerlines and topology relationship, improving both the accuracy and inference speed of lane topology understanding for autonomous driving. Our key innovation lies in reusing intermediate attention resources within distinct transformer decoders. This approach effectively leverages the inherent relational knowledge within the element detection module to enable the modeling of topology relationships among traffic elements and lanes without requiring additional computationally expensive graph networks. Furthermore, we are the first to demonstrate that knowledge can be distilled from models that utilize standard definition (SD) maps to those operates without using SD maps, enabling superior performance even in the absence of SD maps. Extensive experiments on the OpenLane-V2 dataset show that our approach outperforms baseline methods in both accuracy and efficiency, achieving superior results in lane detection, traffic element identification, and topology reasoning. Our code is available at https://github.com/Yang-Li-2000/one-stage.git. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_17617 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Reusing Attention for One-stage Lane Topology Understanding Li, Yang Zhang, Zongzheng Qiu, Xuchong Li, Xinrun Liu, Ziming Wang, Leichen Li, Ruikai Zhu, Zhenxin Gao, Huan-ang Lin, Xiaojian Cui, Zhiyong Zhao, Hang Zhao, Hao Computer Vision and Pattern Recognition Understanding lane toplogy relationships accurately is critical for safe autonomous driving. However, existing two-stage methods suffer from inefficiencies due to error propagations and increased computational overheads. To address these challenges, we propose a one-stage architecture that simultaneously predicts traffic elements, lane centerlines and topology relationship, improving both the accuracy and inference speed of lane topology understanding for autonomous driving. Our key innovation lies in reusing intermediate attention resources within distinct transformer decoders. This approach effectively leverages the inherent relational knowledge within the element detection module to enable the modeling of topology relationships among traffic elements and lanes without requiring additional computationally expensive graph networks. Furthermore, we are the first to demonstrate that knowledge can be distilled from models that utilize standard definition (SD) maps to those operates without using SD maps, enabling superior performance even in the absence of SD maps. Extensive experiments on the OpenLane-V2 dataset show that our approach outperforms baseline methods in both accuracy and efficiency, achieving superior results in lane detection, traffic element identification, and topology reasoning. Our code is available at https://github.com/Yang-Li-2000/one-stage.git. |
| title | Reusing Attention for One-stage Lane Topology Understanding |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2507.17617 |