SPS IVMSP TC Webinar: Deep Learning for Inverse Problems in Imaging

Finding solutions to inverse problems has historically been one of the most important image processing problems; deep learning has provided a tremendous boost to the field, allowing to learn complex models that can be used to restore degraded observations. Indeed, recent deep models can be used to learn good image priors in several different ways, leading to plenty of possible applications even with lack of ground truth data, or lack of training data at all! This talk will focus on recent advances in deep models for restoration, using two application areas as use cases, namely super-resolution and denoising. We will cover models for supervised, self-supervised, unsupervised and one-shot restoration, in the single- and multi-image case, as well as the generation of multiple "good" solutions that are consistent with the observed data. We will show examples of models learning non-local relationships in the data, along with applications to a variety of data types including optical/radar satellite images and point clouds. Finally, we will discuss multimodality as a way to further improve image restoration accuracy.