Guided Super-Resolution for Remote Sensing and Monocular Depth

Guided super-resolution is a powerful technique where a high-resolution guide image aids in enhancing a lower-resolution target image. This talk explores how this principle finds impactful applications in two distinct areas. The first segment features POMELO, a method that adapts guided super-resolution for population mapping, utilizing coarse census data to refine population maps — a significant aid for developing countries. The second part focuses on DADA, a state-of-the-art approach for enhancing depth image resolution. By combining anisotropic diffusion techniques with contemporary neural networks, DADA sets new performance benchmarks in the realm of monocular depth super-resolution.