Researchers have introduced deSEO, a novel methodology for creating paired datasets to address shadow removal in high-resolution satellite imagery. This approach leverages physics-informed techniques and temporal/geometric filtering to derive shadow-free and shadowed image pairs from existing datasets. The developed deSEO model demonstrates improved shadow reduction and structural fidelity, offering a new baseline for satellite Earth observation tasks. AI
Summary written by gemini-2.5-flash-lite from 2 sources. How we write summaries →
IMPACT Provides a new dataset and baseline model for improving satellite image analysis by removing shadows.
RANK_REASON This is a research paper detailing a new methodology and dataset for a specific computer vision task.