Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Improved MambdaBDA Framework for Robust Building Damage Assessment Across Disaster Domains

Alp Eren Gençoğlu, Hazım Kemal Ekenel

Abstract

Reliable post-disaster building damage assessment (BDA) from satellite imagery is hindered by severe class imbalance, background clutter, and domain shift across disaster types and geographies. In this work, we address these problems and explore ways to improve the MambaBDA, the BDA network of ChangeMamba architecture, one of the most successful BDA models. The approach enhances the MambaBDA with three modular components: (i) Focal Loss to mitigate class imbalance damage classification, (ii) lightweight Attention Gates to suppress irrelevant context, and (iii) a compact Alignment Module to spatially warp pre-event features toward post-event content before decoding. We experiment on multiple satellite imagery datasets, including xBD, Pakistan Flooding, Turkey Earthquake, and Ida Hurricane, and conduct in-domain and crossdataset tests. The proposed modular enhancements yield consistent improvements over the baseline model, with 0.8% to 5% performance gains in-domain, and up to 27% on unseen disasters. This indicates that the proposed enhancements are especially beneficial for the generalization capability of the system.

Improved MambdaBDA Framework for Robust Building Damage Assessment Across Disaster Domains

Abstract

Reliable post-disaster building damage assessment (BDA) from satellite imagery is hindered by severe class imbalance, background clutter, and domain shift across disaster types and geographies. In this work, we address these problems and explore ways to improve the MambaBDA, the BDA network of ChangeMamba architecture, one of the most successful BDA models. The approach enhances the MambaBDA with three modular components: (i) Focal Loss to mitigate class imbalance damage classification, (ii) lightweight Attention Gates to suppress irrelevant context, and (iii) a compact Alignment Module to spatially warp pre-event features toward post-event content before decoding. We experiment on multiple satellite imagery datasets, including xBD, Pakistan Flooding, Turkey Earthquake, and Ida Hurricane, and conduct in-domain and crossdataset tests. The proposed modular enhancements yield consistent improvements over the baseline model, with 0.8% to 5% performance gains in-domain, and up to 27% on unseen disasters. This indicates that the proposed enhancements are especially beneficial for the generalization capability of the system.
Paper Structure (28 sections, 4 equations, 2 figures, 11 tables)

This paper contains 28 sections, 4 equations, 2 figures, 11 tables.

Table of Contents

  1. INTRODUCTION
  2. RELATED WORK
  3. BDA Datasets & Benchmarks
  4. Deep Learning in Change Detection
  5. METHODOLOGY
  6. Baseline Model
  7. Enhanced Models
  8. Adopting Focal Loss
  9. Attention Gate Integration
  10. Custom Alignment Module
  11. EXPERIMENTAL SETUP
  12. Datasets
  13. xBD
  14. Pakistan Flooding
  15. Turkey Earthquake
  16. ...and 13 more sections

Figures (2)

  • Figure 1: xBD Sample data: image pairs and mask pairs.
  • Figure 2: Example attention gate outputs. Heatmaps are overlayed on ground truth masks. Attention increases from blue to green to red. The top right heatmap belongs to AGB, the AGs in building decoder, and the bottom right heatmap belongs to AGD, the AGs in damage decoder. AGB mostly focuses on building borders improving performance, while AGD mostly focuses on spaces between the buildings, which does not increase the scores.