Diffusion Distillation for Remote Sensing Salient Object Detection (opens in new tab)

Salient object detection (SOD) in optical remote sensing images (ORSIs) is challenging due to limited spatial details, large-scale variation, and cluttered backgrounds with strong look-alike interference. Although diffusion models can progressively refine object structures and boundaries, their multistep sampling introduces high latency, which limits practical deployment over large areas. This article proposes semantic-guided diffusion distillation network (SGDDiff), which distills a high-cap...