🧪 Test?View on arXiv
Towards Label-Efficient Geospatial Foundation Model Adaptation via Positive-Unlabeled Learning for Building Damage Mapping
Not provided
weak supervisiongeospatial modelingdisaster response
2611.08561
Builder Relevance
16h ago80%
Abstract
This paper presents a framework for adapting geospatial foundation models to efficiently map building damage using minimal labeled data.
Reality Card
Core Claim
The PUMA framework enables effective building damage mapping with only a small number of labeled examples by utilizing weak supervision techniques.
Method / Result
PUMA demonstrated effectiveness in mapping damage during the 2025 Los Angeles wildfires with limited supervision.
Limitations
The reliance on weak supervision may introduce variability in representation learning, affecting reproducibility.
Paper to code
Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.
No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.