Competences

In Video Signal Processing and Transmission cutting-edge algorithms for the acquisition, improvement and reconstruction of images are developed. Besides single-image cameras heterogeneous camera arrays are researched, where individual camera parameters such as spatial resolution, dynamic range, spectral sensitivity or polarization may differ among the involved cameras. In the field of image and video analysis approaches from the field of machine learning are mainly used. One possible area of application is inspection technology. Furthermore, the chair researches on the latest issues in image and video coding. Among others, special adjustments and enhancements of already existing coding standards for special cases such as medical data, screen content or fish-eye video sequences are examined.

In the field of Audio and Acoustic Signal Processing model- and data-based algorithms are researched for acoustic human/machine-interaction and for the analysis and playback of real and virtual acoustic scenes. These involve different application, e.g., telecommunication and telepresence, hearing aid devices, smart homes, human/machine-communication or immersive multimedia environments. As a key characteristic, the usage of the spatial dimension via multi-channel signal processing plays a decisive role in many projects.

In machine learning, and especially in deep learning, the focus is on both fundamental and applied research. Machine learning algorithms are developed for anomaly detection, uncertainty estimation, out-of-distribution detection, few-shot learning, noisy label learning, as well as hardware motivated problems such as model compression and efficient design of neural network architectures. Applications range from automated driving, computer vision and medical image analysis to signal processing. For example, trajectory and motion prediction, localisation, and image and point cloud segmentation are some examples of practical problems.