FLoRA: Fusion-Latent for Optical Reconstruction and Flood Area Segmentation via Cross-Modal Multitask Distillation Network (opens in new tab)

Accurate flood-water mapping is critical for disaster management, yet current methods struggle to fully exploit the potential of spaceborne imagery. Optical data offer high interpretability but is limited by environmental conditions, whereas synthetic aperture radar (SAR) provides reliable all-weather coverage with reduced visual interpretability. Fusion-latent for optical reconstruction and flood area segmentation (FLoRA) is a cross-modal multitask framework that jointly reconstructs high-fi...