HRHD-HK
收藏资源简介:
HRHD-HK是由香港大学建筑学院创建的一个专注于高密度和高楼层城市场景的3D语义分割基准数据集。该数据集包含2.73亿个彩色摄影测量3D点,分布在150个瓦片中,覆盖香港的多样化城市环境。数据集的语义标签包括建筑、植被、道路、水体、设施、地形和车辆。HRHD-HK旨在通过引入亚洲高密度和高楼层城市场景以及亚热带自然景观,补充现有的基准数据集。数据集的创建过程涉及数据采集、数据整理、语义标注和训练-验证-测试分割四个步骤。HRHD-HK的应用领域包括智能城市应用,如机器人、自动驾驶和导航以及城市分析,旨在解决城市对象功能和形态的理解问题。
HRHD-HK is a 3D semantic segmentation benchmark dataset created by the Faculty of Architecture, The University of Hong Kong, focusing on high-density and high-rise urban scenarios. This dataset contains 273 million colored photogrammetric 3D points, distributed across 150 tiles, covering the diverse urban environments of Hong Kong. Its semantic labels include buildings, vegetation, roads, water bodies, facilities, terrain, and vehicles. HRHD-HK aims to complement existing benchmark datasets by introducing Asian high-density and high-rise urban scenarios as well as subtropical natural landscapes. The creation of this dataset involves four steps: data acquisition, data organization, semantic annotation, and train-validation-test split. The application scenarios of HRHD-HK cover smart city applications such as robotics, autonomous driving, navigation, and urban analysis, aiming to address the challenge of understanding the functions and morphologies of urban objects.




