Perception of People for HCI

Highlights

  • Person detection and tracking
  • Body pose estimation
  • Gesture and activity recognition
  • Face recognition and analysis
  • Gaze and head pose estimation
Selected Publications
Author Title Source

A. Roitberg, M. Haurilet, M. Martinez, R. Stiefelhagen.

International Conference on Pattern Recognition (ICPR), IEEE, January 2021, (oral presentation), pdf

M. Martin*, A. Roitberg*, M. Haurilet, M. Horne, S. Reiß, M. Voit, R. Stiefelhagen

International Conference on Computer Vision (ICCV), Seoul, South Korea, Oct. 2019 (*equal contribution, pdf, bib)

A. Roitberg, Z. Al-Halah, R. Stiefelhagen

British Machine Vision Conference (BMVC), Newcastle upon Tyne, UK

A. Schwarz, M. Haurilet, R. Stiefelhagen

IEEE Conference on Computer Vision and Pattern Recognition (CVPR Workshop on Computer Vision in Vehicle Technology and Autonomous Driving Challenge), Honolulu, HI, USA, July 2017

M. S. Sarfraz, R. Stiefelhagen

In Proceedings British Machine Vision Conference (BMVC) September 2015 (Best Industry Paper Award)

Body Pose, Tracking, Activity Recognition

Another focus of research is in developing perceiving interactive environments (e.g. Smart Rooms), tracking persons and their activities for a context-adapted and customized support. Ongoing work in this application context include the detection of persons, head pose tracking, action recognition. Some related applications currently investigated are interactive operating environments (cooperation with the University of Heidelberg) and human detection for driver assistance systems (several ongoing external promotions).

Face Recognition and Analysis

Face recognition plays an important role in perception components for HCI. Our research is focused on addressing the problem of person identification in realistic imaging conditions, e.g. surveillance video feeds etc. Matching a person's captured facial image in these conditions to a stored high resolution database image requires investigating new methods. Here most of our research addresses the problems of face recognition and analysis including face recognition across large pose differences, lighting, image resolution and different sensor modalities.

We also investigate automatic facial analysis for facial expressions and emotion recognition.