SGMA: Semantic-Guided Modality-Aware Segmentation for Remote Sensing With Incomplete Multimodal Data (opens in new tab)

Multimodal semantic segmentation (MSS) integrates complementary information from diverse sensors for remote sensing Earth observation. However, practical systems often encounter missing modalities due to sensor failures or incomplete coverage, termed incomplete MSS (IMSS). IMSS faces three key challenges: 1) multimodal imbalance, where dominant modalities (e.g., RGB) suppress fragile ones (e.g., DSM, NIR, and SAR); 2) intraclass variation in scale, shape, and orientation across modalities; an...