全球道路表面特征数据集
收藏资源简介:
全球道路表面特征数据集是由海德堡地理信息技术研究所 (HeiGIT) 基于Mapillary街景图像创建的,包含104,523,781张图像,覆盖全球范围。数据集通过深度学习方法,结合SWIN-Transformer和CLIP-and-DL分割技术,对道路表面进行分类(铺砌或未铺砌)。数据集的创建过程包括图像下载、预处理、模型训练和数据匹配等步骤。该数据集主要应用于城市规划、灾害路线优化、物流优化等领域,旨在解决全球道路表面数据不完整的问题,支持可持续发展目标(SDGs)。
The Global Road Surface Feature Dataset was created by the Heidelberg Institute for Geoinformation Technology (HeiGIT) based on Mapillary street view images. It contains 104,523,781 images with global coverage. The dataset adopts deep learning methods combined with SWIN-Transformer and CLIP-and-DL segmentation technologies to classify road surfaces into paved and unpaved categories. The dataset creation process includes steps such as image downloading, preprocessing, model training, and data matching. This dataset is mainly applied in fields such as urban planning, disaster route optimization, and logistics optimization. It aims to address the problem of incomplete global road surface data and support the Sustainable Development Goals (SDGs).




