What question did this study set out to answer?

The aim is to develop a deep learning framework for rapid building damage detection using multimodal satellite data after earthquakes.

July 11, 2026Open Access

A Deep Learning Framework for Rapid Building Damage Detection through Multimodal Data Fusion: Application to the 2025 Myanmar Earthquake

LRLuigi RussoUniversity of Pavia DTDeodato TapeteAgenzia Spaziale Italiana SUSilvia Liberata UlloUniversity of Sannio

Key Points

The aim is to develop a deep learning framework for rapid building damage detection using multimodal satellite data after earthquakes.
Formulated building damage detection as a binary image classification task using a unified ResNet-18 backbone.
Integrated data from COSMO-SkyMed SAR imagery, Maxar optical data, OpenStreetMap, and UNOSAT annotations.
Evaluated three fusion methods: Early Fusion, Late Fusion, and a novel Middle Fusion approach.
Middle Fusion outperformed Early Fusion and Late Fusion with higher precision and F1-score.
Achieved robust performance across various modality configurations in five-fold cross-validation.
Demonstrated effective mapping of building damage through combined spatial contexts from SAR, optical data, and building footprints.

Abstract

Abstract. Rapid and reliable assessment of building damage after major earthquakes is essential for effective emergency response and recovery planning. This study formulates post-disaster building damage detection (BDD) as a binary image classification task (damaged vs. undamaged buildings) using multimodal satellite data and a unified ResNet-18 backbone to enable a controlled comparison of fusion strategies. The analysis focuses on the Mw 7.7 Myanmar earthquake of 28 March 2025 and integrates post-event COSMO-SkyMed Second Generation (CSG) dual-polarization (HH, HV) SAR imagery, Maxar optical data, OpenStreetMap (OSM) building footprints, and UNOSAT damage annotations. Three fusion paradigms are evaluated: Early Fusion (EF), Late Fusion (LF), and a novel Middle Fusion (MF) approach. The proposed MF framework introduces a Footprint-Guided Cross-Attention (FGCA) mechanism that uses building geometry as a spatial prior to guide feature-level interaction between SAR and optical representations. Five-fold cross-validation results show that MF consistently outperforms EF and LF, achieving higher precision, F1-score, and robustness across modality configurations. By jointly exploiting SAR structural sensitivity, optical detail, and footprint-based spatial context, the proposed Footprint-Guided Middle Fusion (FGMF) framework enables accurate and scalable building damage mapping from heterogeneous Earth Observation (EO) data.

Ask AI

Helpful

Bookmark

View Full Paper