面向卫星图像道路图提取:一个全局规模数据集和一种新方法
计算机视觉与模式识别
2024-11-27 v1
摘要
最近,由于其在自动驾驶、导航等领域的关键作用,道路图提取正受到日益关注。然而,准确且高效地提取道路图始终是一个持久的挑战,主要由于标记数据的严重稀缺。为此,我们收集了一个全局规模的卫星道路图提取数据集,即Global-Scale数据集。具体而言,Global-Scale数据集是现有最大公开道路提取数据集的约20倍,覆盖全球13,800平方公里。此外,我们开发了一种新型道路图提取模型,即SAM-Road++,该模型采用节点引导的重采样方法,以缓解SAM-Road——一种开创性的最先进道路图提取模型——在训练和推理之间的不匹配问题。 Furthermore, we propose a simple yet effective ``extended-line'' strategy in SAM-Road++ to mitigate the occlusion issue on the road. Extensive experiments demonstrate the validity of the collected Global-Scale dataset and the proposed SAM-Road++ method, particularly highlighting its superior predictive power in unseen regions. The dataset and code are available at \url{https://github.com/earth-insights/samroadplus}.。
引用
@article{arxiv.2411.16733,
title = {Towards Satellite Image Road Graph Extraction: A Global-Scale Dataset and A Novel Method},
author = {Pan Yin and Kaiyu Li and Xiangyong Cao and Jing Yao and Lei Liu and Xueru Bai and Feng Zhou and Deyu Meng},
journal= {arXiv preprint arXiv:2411.16733},
year = {2024}
}