BRIGHT: A globally distributed multimodal building damage assessment dataset with very-high-resolution for all-weather disaster response

TL;DR

BRIGHT tackles the challenge of all-weather building damage assessment by introducing a globally distributed, open multimodal dataset that pairs pre-event optical imagery with post-event SAR data at sub-meter resolution across 14 disasters and 23 regions. It benchmarks a spectrum of models, including end-to-end and localization–classification pipelines, and probes cross-event generalization, unsupervised domain adaptation, and unsupervised multimodal tasks (UMCD, UMIM). The results highlight strong baselines, show that pre-event optical data enriches damage discrimination beyond localization, and demonstrate that cross-event transfer remains challenging, motivating robust domain-agnostic methods and richer multimodal extensions. BRIGHT thus provides a substantial resource for developing all-weather disaster-response AI, supports future EO foundation models, and invites expansion to additional modalities like fully polarimetric SAR and LiDAR for deeper structural understanding.

Abstract

Disaster events occur around the world and cause significant damage to human life and property. Earth observation (EO) data enables rapid and comprehensive building damage assessment (BDA), an essential capability in the aftermath of a disaster to reduce human casualties and to inform disaster relief efforts. Recent research focuses on the development of AI models to achieve accurate mapping of unseen disaster events, mostly using optical EO data. However, solutions based on optical data are limited to clear skies and daylight hours, preventing a prompt response to disasters. Integrating multimodal (MM) EO data, particularly the combination of optical and SAR imagery, makes it possible to provide all-weather, day-and-night disaster responses. Despite this potential, the development of robust multimodal AI models has been constrained by the lack of suitable benchmark datasets. In this paper, we present a BDA dataset using veRy-hIGH-resoluTion optical and SAR imagery (BRIGHT) to support AI-based all-weather disaster response. To the best of our knowledge, BRIGHT is the first open-access, globally distributed, event-diverse MM dataset specifically curated to support AI-based disaster response. It covers five types of natural disasters and two types of man-made disasters across 14 regions worldwide, with a particular focus on developing countries where external assistance is most needed. The optical and SAR imagery in BRIGHT, with a spatial resolution between 0.3-1 meters, provides detailed representations of individual buildings, making it ideal for precise BDA. In our experiments, we have tested seven advanced AI models trained with our BRIGHT to validate the transferability and robustness. The dataset and code are available at https://github.com/ChenHongruixuan/BRIGHT. BRIGHT also serves as the official dataset for the 2025 IEEE GRSS Data Fusion Contest.

Figures (12)

• Figure 1: An example of the wildfire occurring in Maui, Hawaii, USA, August 2023. (a) Pre-event optical imagery (© Maxar). (b) Post-event optical image (© Maxar) with land-cover features obscured by wildfire smoke. (c) Post-event SAR imagery (© Capella Space) unaffected by smoke, showing the disaster area.
• Figure 2: Geographic distribution of disaster events present in Bright.
• Figure 3: Overall flowchart of developing the Bright dataset.
• Figure 4: Thumbnails of local areas in 14 disaster events in the Bright dataset. The sources of EO images are illustrated in Table \ref{['tbl:BRIGHT_info']}. For visualization purposes, different events have different scales.
• Figure 5: Statistics of the Bright dataset. (a) Distributions of band values of samples from four study sites. (b) Distribution of building scales. (c) Feature distribution of buildings of four events under two imaging modalities. (d) Percentage of building and background pixels and percentage of different damage levels in building pixels.