BRIGHT
收藏资源简介:
BRIGHT数据集是一个全球分布的多模态建筑损伤评估数据集,由东京大学、RIKEN、苏黎世联邦理工学院和微软亚洲研究院联合创建。该数据集包含高分辨率的光学和SAR图像,空间分辨率在0.3至1米之间,覆盖了12个地区的5种自然灾害和2种人为灾害,特别关注发展中国家。数据集包含超过35万条建筑实例,提供了详细的建筑损伤信息,适用于精确的损伤评估。数据集的创建过程包括从多个灾害事件中收集光学和SAR图像,并进行多级标注,区分受损和完全毁坏的建筑。BRIGHT数据集的应用领域包括灾害响应、建筑损伤评估和人工智能模型的训练与评估,旨在通过全天候的灾害响应减少人员伤亡和财产损失。
The BRIGHT dataset is a globally distributed multimodal building damage assessment dataset jointly created by The University of Tokyo, RIKEN, ETH Zurich, and Microsoft Research Asia. This dataset contains high-resolution optical and SAR images with spatial resolutions ranging from 0.3 to 1 meter, covering 5 types of natural disasters and 2 types of man-made disasters across 12 regions, with a particular focus on developing countries. It includes more than 350,000 building instances, providing detailed building damage information for accurate damage assessment. The dataset development process involves collecting optical and SAR images from multiple disaster events and conducting multi-level annotations to distinguish between damaged and completely destroyed buildings. Application scenarios of the BRIGHT dataset include disaster response, building damage assessment, as well as the training and evaluation of artificial intelligence models, aiming to reduce casualties and property losses through all-weather disaster response.




