"Phy-OSShip-MS: A Physics-guided Optical-SAR Ship Imagery Dataset with Multi-Source Attributes"

"High-quality paired optical and Synthetic Aperture Radar (SAR) datasets are fundamental for cross-modal ship detection. However, existing fusion datasets, typically derived from historical archives, suffer from prolonged revisit intervals. This results in geographically aligned backgrounds with spatiotemporally missing targets, where highly maneuverable ships appear in only one modality, rendering such data unsuitable for learning synchronous target characteristics. To address this scarcity and the lack of complex maritime environments, this repository hosts the Phy-OSShip-MS Dataset. This is a large-scale, strictly paired Optical-SAR ship detection dataset generated via a novel \"Hierarchical Physics-Guided Style Transfer\" framework, providing Instance-level spatial consistency for robust cross-modal model training."