LADAR: Language-Adaptive Dynamic Aqua Refinement for Underwater Image Enhancement (opens in new tab)

Quality degradation in underwater images poses significant challenges for oceanic remote sensing and marine applications. The inherent properties of the water column, such as absorption and scattering, introduce severe distortions that corrupt optical data, impeding critical remote sensing tasks, like seafloor mapping, resource monitoring, and environmental surveillance. Existing deep learning methods for underwater image enhancement (UIE) often lack high-level semantic understanding, creatin...