Project Objective

Computer vision systems process enormous amounts of data, and many of their tasks can be formulated as combinatorial optimization problems. Nevertheless, quantum computing has hardly been used in this field to date. The project investigates how computer vision can benefit from quantum hardware. It pursues two lines of research across five work packages: First, vision tasks are formulated as QUBO problems so that quantum annealers can solve them. This includes embedding constraints, learned cost functions, and the training of binary neural networks. Second, quantum autoencoders for 3D data are being developed on gate-based quantum computers, with the aim of enabling more compact and expressive representations.

Our Mission

Two departments are collaborating in Siegen: the Department of Computer Vision (Prof. Dr. Michael Möller), with its expertise in optimization and machine learning, and the Department of Experimental Quantum Optics (Prof. Dr. Christof Wunderlich), which brings the perspective of quantum hardware. Our partner is the “4D Quantum Computer Vision” research group led by Dr.-Ing. Vladislav Golyanik at the Max Planck Institute for Informatics in Saarbrücken.