Researchers have introduced ShadowTransfer, a new benchmark designed to evaluate how well shadow detection models perform when trained in one geographic region and applied in another. The benchmark includes 4,500 human-verified shadow masks across three U.S. cities and uses a leave-one-city-out evaluation protocol. Initial tests show significant performance drops when models are transferred to new regions, with some adaptation methods even degrading performance further. The study also proposes a diagnostic framework and a plug-in module called Spectral Information Bottleneck (SIB) to address these transfer failures. AI
IMPACT Highlights challenges in geographic transfer for computer vision models, potentially impacting deployment of AI systems in diverse real-world environments.
RANK_REASON The cluster describes a new benchmark and research paper published on arXiv. [lever_c_demoted from research: ic=1 ai=1.0]
- arXiv
- Institut National de Recherche en Informatique et en Automatique
- ShadowTransfer
- Spectral Information Bottleneck
- U.S.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →