Topo-boundary: A Benchmark Dataset on Topological Road-boundary Detection Using Aerial Images for Autonomous Driving

This paper is accepted by IEEE Robotics and Automation Letters(RA-L) and The IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2021, which is available atarxiv,IEEE.

Abstract

Road-boundary detection is important for autonomous driving. For example, it can be used to constrain vehicles running on road areas, which ensures driving safety. Compared with online road-boundary detection using on-vehicle cameras/Lidars, offline detection using aerial images could alleviate the severe occlusion issue. Moreover, the offline detection results can be directly used to annotate high-definition (HD) maps. In recent years, deep-learning technologies have been used in offline detection. But there is still lacking a publicly available dataset for this task, which hinders the research progress in this area. So in this paper, we propose a new benchmark dataset, named Topo-boundary, for offline topological road-boundary detection. The dataset contains 25,295 1000*1000-sized 4-channel aerial images. Each image is provided with 8 training labels for different sub-tasks. We also design a new entropy-based metric for connectivity evaluation, which could better handle noises or outliers. We implement and evaluate 3 segmentation-based baselines and 5 graph-based baselines using the dataset. We also propose a new imitation-learning-based baseline which is enhanced from our previous work. The superiority of our enhancement is demonstrated from the comparison.

Dataset description

This dataset is for topological road-boundary detection in aerial images for autonomous driving purposes. We provide 25,295 high-resolutin aerial images and each image have multiple labels for different learining tasks. We also provide 9 baseline models.

This dataset is based on the GIS database NYC Planimetric Database.

Supplementary

Supplementary document provides more details about the evaluation metrics and data structure.

Code and data

Please check our github repo and follow the steps.

Contact

For any questions, please send email to zxubg at connect dot ust dot hk.

Citation

@article{xu2021topo,
  title={Topo-boundary: A benchmark dataset on topological road-boundary detection using aerial images for autonomous driving},
  author={Xu, Zhenhua and Sun, Yuxiang and Liu, Ming},
  journal={IEEE Robotics and Automation Letters},
  volume={6},
  number={4},
  pages={7248--7255},
  year={2021},
  publisher={IEEE}
}

@article{xu2021icurb,
  title={iCurb: Imitation Learning-Based Detection of Road Curbs Using Aerial Images for Autonomous Driving},
  author={Xu, Zhenhua and Sun, Yuxiang and Liu, Ming},
  journal={IEEE Robotics and Automation Letters},
  volume={6},
  number={2},
  pages={1097--1104},
  year={2021},
  publisher={IEEE}
}