David Schneider

Research Assistant at CV:HCI. PhD Student. Computer Vision. Machine Learning.

prof_pic.jpg

Changing environements are a challenge for modern deep learning based human action recognition. I focus on robust, domain invariant systems which observe humans during daily and non-daily actions and recognize their behaviour in order to improve robotic assistance systems for activities of daily living.

In this field of research I am less interested in the performance within a specific datasets, but rather the performance across datasets: How much worse is the recognition accuracy under different lighting conditions, background sceneries or even slightly changing semantics of certain labels.

I finished my M.Sc. in Computer Science in 2021 and started my PhD at the CV:HCI Lab afterwards, both at the Karlsruhe Institute of Technology (KIT). I am involved in the JuBot Project which develops robotic systems to support elderly people in everyday living situtations and which is funded by the Carl-Zeiss-Stiftung.

selected publications

  1. Pose-Based Contrastive Learning for Domain Agnostic Activity Representations
    Schneider, David, Sarfraz, Saquib, Roitberg, Alina, and Stiefelhagen, Rainer
    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops Jun 2022
  2. Let’s Play for Action: Recognizing Activities of Daily Living by Learning from Life Simulation Video Games
    Roitberg, Alina*,  Schneider, David*, Djamal, Aulia, Seibold, Constantin, Reiß, Simon, and Stiefelhagen, Rainer
    In 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) Jun 2021