Deep Learning for Computer Vision (Dagstuhl Seminar 17391)

The field of computer vision engages in the goal to enable and enhance a machine’s ability to infer knowledge and information from spatial and visual input data. Recent advances in data-driven learning approaches, accelerated by increasing parallel computing power and a ubiquitous availability of large amounts of data, pushed the boundaries of almost every vision related subdomain. The most prominent example of these machine learning approaches is a so called deep neural network (DNN), which works as a general function approximator and can be trained to learn a mapping between given input and target output data. Research on and with these DNN is generally referred to as Deep Learning. Despite its high dimensional and complex input space, research in the field of computer vision was and still is one of the main driving forces for new development in machine and deep learning, and vice versa. This seminar aims to discuss recent works on theoretical and practical advances in the field of deep learning itself as well as state-of-the-art results achieved by applying learning based approaches to various vision problems. Our diverse spectrum of topics includes theoretical and mathematical insights focusing on a better understanding of the fundamental concepts behind deep learning and a multitude of specific research topics facilitating an exchange of knowledge between peers of the research community.