地震损毁房屋多尺度遥感特征数据集

地震损毁房屋多尺度遥感特征数据集涵盖了2010-2023年期间发生的五次重大地震灾害的高分辨率灾前与灾后光学遥感影像数据，并基于遥感影像的光学解译特征，将受损建筑分为三类，制作了受损严重建筑的标签。
地震损毁房屋多尺度遥感特征数据集在研究地震灾害的影响、形成机制以及灾后恢复过程中具有不可或缺的作用。由于不同地区在不同时间所面临的地震震源、震中位置、地形和建筑结构等因素存在显著差异，导致地震损毁房屋的识别和评估精度常常受到多种限制。完整且准确地识别地震损毁房屋不仅有助于理解地震的时空分布和影响模式，还能分析地震损毁的驱动因素，如震级、震中距离、地质条件等，评估地震对人类、财产和自然环境的多重影响。此数据集综合了不同尺度的灾前和灾后遥感数据以及相应的建筑损毁标签，填补了我国在地震损毁建筑数据集领域的空白。通过该数据集，能够为相关地震损毁建筑检测相关模型的训练与泛化提供重要支持，并为地震应急响应和减灾策略的制定提供科学依据。同时，该数据集提供的初级影像特征，可有效促进地震损毁建筑检测算法的提升与应用，从而加速我国在地震灾害管理和决策领域的数据驱动与算法自动化进程。

The Multi-scale Remote Sensing Feature Dataset of Earthquake-damaged Houses includes high-resolution pre- and post-disaster optical remote sensing imagery of five major earthquakes that occurred between 2010 and 2023. Based on the optical interpretation features of remote sensing images, damaged buildings are classified into three categories, and labels for severely damaged buildings are generated.

This dataset plays an indispensable role in studying the impacts, formation mechanisms, and post-disaster recovery processes of earthquake disasters. Significant differences exist in factors such as earthquake seismic sources, epicenter locations, terrain, and building structures across different regions and time periods, which often impose multiple constraints on the accuracy of earthquake-damaged building identification and assessment. Accurately and comprehensively identifying earthquake-damaged buildings not only helps to understand the spatial-temporal distribution and impact patterns of earthquakes, but also enables the analysis of driving factors of earthquake damage including magnitude, epicentral distance, geological conditions, etc., as well as the evaluation of multiple impacts of earthquakes on humans, property, and the natural environment.

This dataset integrates multi-scale pre- and post-disaster remote sensing data and corresponding building damage labels, filling the gap in the field of earthquake-damaged building datasets in China. It can provide critical support for the training and generalization of relevant earthquake-damaged building detection models, and offer scientific foundations for earthquake emergency response and the formulation of disaster mitigation strategies. Meanwhile, the primary image features provided by this dataset can effectively facilitate the improvement and application of earthquake-damaged building detection algorithms, thereby accelerating the data-driven and algorithm automation process in the field of earthquake disaster management and decision-making in China.