Objective: Visual perception is a key ability for an autonomous agent to assess the state of its environment. The research of the Computer Vision Group in Freiburg aims to improve systems for visual understanding; to bridge the gap to human learning by achieving generalization in open-world and out-of-distribution settings; and to reduce the need for human supervision by using self- or weakly supervised learning. Research topics intersect with other areas including robotics, natural language processing and bioinformatics. Research applications include optical flow estimation (FlowNet) and biomedical image segmentation (U-Net).

Research Areas: 3D Scene Understanding, Vision-Language Modelling, Biomedical Image Analysis, Generative Vision, Vision for Robotics, Robustness, Optical Flow, Visual Correspondence, Uncertainty Estimation