Seminar by Julien Mairal (INRIA)

Abstract

Deep learning has revolutionized image processing and is often considered to outperform classical approaches based on accurate modeling of the image formation process. In this presentation, we will discuss the interplay between model-based and learning-based paradigms, and show that hybrid approaches hold great promise for scientific imaging, where interpretability and robustness to real-world degradation are important. We will present two applications involving super-resolution and high-dynamic-range imaging, as well as exoplanet detection via direct imaging at high contrast.