EG-CMFNet: An Edge-Guided Cross-Modal Fusion Network for Remote Sensing Semantic Segmentation (opens in new tab)

Multimodal remote sensing data often exhibit complementary information about ground objects and are widely utilized for semantic segmentation tasks. However, spatial heterogeneity and misalignment from different modalities hinder the utilization of multimodal information. To address these challenges, an edge-guided cross-modal fusion network (EG-CMFNet) is proposed for multimodal remote sensing imagery. We introduce a multimodal adaptive feature interaction (MAFI) module to adaptively interac...