Damage-TriageFormer: A Foundation-Model Framework for Typology-Based Building Damage Assessment from Mono-Temporal Imagery

2026-06-10

AI Summary

This study addresses the limitations of existing building damage assessment methods, which either require pre- and post-disaster image pairs or yield only coarse-grained damage levels, thereby hindering precise emergency response. To overcome this, the authors propose a fine-grained damage type recognition framework that operates on a single post-disaster image and incorporates building footprint priors to distinguish between roof versus structural damage and partial versus total collapse, defining five distinct damage categories. The work introduces DamageTriage-Bench, the first benchmark dataset tailored for damage-type classification, and presents an end-to-end model built upon a DINOv3 ViT-L backbone, enhanced with a Simple Feature Pyramid to improve instance pooling resolution, alongside a two-stage gated damage head and a severity regression objective. The model achieves macro F1 scores of 0.624 and 0.619 on validation and independent test sets, with F1 scores of 0.91 and 0.84 for the undamaged and completely collapsed classes, respectively.

Abstract

Decision-relevant building damage assessment is critical for prioritizing resources and recovery after a disaster, yet most automated methods either flatten damage into a single severity scale (no damage, minor, major, destroyed) or require paired pre- and post-event imagery that is often unavailable for emerging hazards. This paper presents Damage-TriageFormer, a single-image, post-event, footprint-conditioned model that produces a damage typology rather than a severity scale. We contribute: (1) DamageTriage-Bench, a new benchmark built from NOAA Emergency Response Imagery across Hurricane Michael (2018), Hurricane Helene (2024), and the 2025 Los Angeles wildfire complex, with five typology classes that distinguish roof damage from structural damage and, within each, partial from total extent; and (2) Damage-TriageFormer, which extends a DINOv3 ViT-L backbone with a Simple Feature Pyramid for higher-resolution instance pooling, a two-stage gated damage head, and an auxiliary severity-regression objective. Our model achieves macro F1 of 0.624 on validation and 0.619 on a held-out stratified test set, performing strongest where operational triage needs it most, with per-class F1 of 0.91 and 0.84 on undamaged buildings and total structural collapse, respectively. While the rare Total Roof Damage class remains difficult due to its limited examples and an inherently ambiguous label boundary, our results show that single-image post-event imagery can support actionable building damage typing, enabling targeted emergency response and resource allocation without a pre-event reference.

Problem

• building damage assessment
• damage typology
• mono-temporal imagery
• disaster response
• emergency triage

Innovation

• damage typology
• mono-temporal imagery
• foundation model
• instance-aware pooling
• gated damage head