PriorCLIP: Visual-Prior-Guided Vision–Language Model for Remote Sensing Image–Text Retrieval
Remote sensing image–text retrieval (RSITR) plays a crucial role in remote sensing interpretation, yet remains challenging under both closed-domain and open-domain scenarios due to semantic noise and domain shifts. To address these challenges within a coherent framework, we propose PriorCLIP, a visual-prior-guided visi...