We are working in the following fields, addressing the challenges of trustworthy and explainable machine learning methods:
Motion Analysis:
- Automated Motion Analysis: based on motion data like Inertial Measurement Units (IMU), skeleton data from 3D-Depth cameras. The overall goal is to build systems which are able to perform motion analysis with user feedback with low-cost sensors. Using IMUs, those approaches are even completely mobile and scalable to many simultaneous users.
- Sports climbing and belaying: improving climbing and belaying technique using mobile sensors and automatic assessment of the performance.
Image analysis:
- Improving document quality as an integral part of automated document classification and managing. Large language models are applied together with image-enhancement models to improve the text-understanding.
- Automated analysis of all kind of image data (for example shearography images), with a modular and explainable model architecture.