Building Damage Assessment
Overview
Building damage assessment studies how to identify buildings and grade their damage from aerial or satellite images, often comparing pre- and post-disaster views but sometimes using post-event imagery alone. Common approaches include CNNs and vision transformers for detection or segmentation, paired-image change detection, and domain adaptation to transfer models across disasters; newer systems also use foundation models and vision-language models. Benchmarks such as xBD and RescueNet measure building-level detection and damage classification, while some evaluations test transfer to unseen events or regions. A central challenge is that models can distinguish obvious damage yet struggle with subtle severity levels, image misalignment, and shifts in sensor or disaster conditions, making robust cross-event evaluation important.
Latest papers
25 papers4 papers in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.