Damage-TriageFormer extends a DINOv3 ViT-L backbone with a feature pyramid and gated head to classify mono-temporal post-event imagery into five damage typologies on a new NOAA-derived benchmark, reporting macro F1 of 0.62.
Building damage detection using satellite images and patch-based transformer methods.arXiv preprint arXiv:2602.08117, 2026
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Damage-TriageFormer: A Foundation-Model Framework for Typology-Based Building Damage Assessment from Mono-Temporal Imagery
Damage-TriageFormer extends a DINOv3 ViT-L backbone with a feature pyramid and gated head to classify mono-temporal post-event imagery into five damage typologies on a new NOAA-derived benchmark, reporting macro F1 of 0.62.