Computer Vision & Graphics

The Computer Vision & Graphics (CVG) group conducts cutting-edge research at the intersection of computer vision, computer graphics, and artificial intelligence. Our research aims at capturing, understanding, and digitally representing the visual world from images and video to complex 3D and dynamic environments.

We develop novel AI-based methods for visual perception, 3D scene understanding and reconstruction, modelling, synthesis, and visualization. A particular focus lies on combining modern machine learning with geometric, physical, and domain knowledge to create models that are accurate, efficient, adaptable, and robust under real-world conditions. By closely linking analysis and synthesis, we develop methods that not only interpret visual observations but also build rich digital representations of objects, humans, and environments.

Our research spans fundamental methods in computer vision and visual computing as well as their transfer into practical solutions. Current topics include 2D and 3D scene understanding, neural scene representations, human modelling and animation, generative models, computational imaging, biomedical image analysis, and augmented and extended reality.

Working closely with partners from industry and research, we translate these technologies into applications in fields such as industrial production, medicine, multimedia, and security.