Expanding Horizons: U-Net Enhancements for Semantic Segmentation, Forecasting, and Super-Resolution in Ocean Remote Sensing
Haoyu Wang, Xiaofeng Li
Abstract
Originally designed for medical segmentation, the U-Net model excels in ocean remote sensing for segmentation, forecasting, and image enhancement. We propose enhancements like attention mechanisms, knowledge-data integration, and diffusion models to improve small target detection, ocean phenomena forecasting, and image super-resolution, expanding U-Net’s application and support in oceanographic research.
Topics & Concepts
SegmentationRemote sensingComputer scienceImage segmentationNet (polyhedron)High resolutionImage (mathematics)DiffusionArtificial intelligenceMeteorologyEnvironmental scienceGeologyGeographyMathematicsPhysicsGeometryThermodynamicsSeismic Imaging and Inversion TechniquesUnderwater Acoustics ResearchOceanographic and Atmospheric Processes