2026-10-12T14:14:16Z https://www.scidb.cn/oai

10.57760/sciencedb.0109o 2026-08-31T10:26:07Z

2026-08-31 D³-Dataset: A Multi-Temporal Fine-Grained Damage Assessment Dataset for Critical Infrastructure doi:10.57760/sciencedb.0109o en The D³-Dataset is developed for multi-temporal fine-grained damage assessment of critical infrastructure, covering airports, ports, and industrial facilities affected by different disaster events. It provides object-level annotations for 15 infrastructure categories with four damage levels: intact, minor damage, major damage, and destroyed. The released dataset package includes polygon-based annotation files, single-channel segmentation masks, detailed scene-level metadata, and information required to identify and reconstruct the corresponding multi-temporal imagery. Due to redistribution restrictions associated with the source imagery, the original Google Earth imagery is not redistributed. Detailed acquisition metadata and data preparation information are provided to facilitate reproducibility and further research on critical infrastructure localization and damage assessment. Remote Sensing; Damage Assessment; Multi-Temporal Imagery; Change Detection; Semantic Segmentation 张伊丹 王天琦 冯瑛超 张相皓 刁文辉 李宁 刘煜 张驰 PUBLIC https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank