SensatUrban
收藏资源简介:
SensatUrban是一个大规模的城市尺度3D点云数据集,由牛津大学创建。该数据集包含来自英国三个城市的近30亿个丰富标注的点,覆盖约7.6平方公里的城市景观。每个3D点都被标注为13个语义类别之一,如地面、植被、建筑等。数据集通过高分辨率无人机摄影测量系统生成,提供了独特的俯视和斜视视角,以及自然颜色信息,适用于智能城市规划和管理等应用。数据集的创建过程涉及从空中图像重建3D点云,并通过专业工具手动标注语义标签。SensatUrban旨在解决大规模城市环境中3D点云的精细语义理解问题,为深度学习算法提供丰富的训练和测试资源。
SensatUrban is a large-scale urban 3D point cloud dataset created by the University of Oxford. This dataset contains nearly 3 billion richly annotated points from three cities in the United Kingdom, covering an urban landscape of approximately 7.6 square kilometers. Each 3D point is annotated with one of 13 semantic categories, such as ground, vegetation, building, and others. The dataset is generated using a high-resolution unmanned aerial vehicle (UAV) photogrammetry system, which offers unique nadir and oblique viewing angles as well as natural color information, making it suitable for applications such as smart city planning and management. The dataset creation process involves reconstructing 3D point clouds from aerial imagery and manually annotating semantic labels with professional tools. SensatUrban aims to address the problem of fine-grained semantic understanding of 3D point clouds in large-scale urban environments, providing abundant training and testing resources for deep learning algorithms.




