home | photography | video | research | SEEDS superpixels
 
 
 


Michael Van den Bergh

Michael Van den Bergh obtained his Master of Science in Electrical Engineering degree from the KU Leuven in 2004. He completed his master thesis "Perceptive User Interface" (video) at the Center for Processing Speech and Images (PSI) at the Electrical Engineering institute (ESAT). He joined the Computer Vision Laboratory at ETH Zurich in 2005 where he worked as a research assistant and Ph.D. student in the field of Human-computer Interaction. In 2010, he finished his doctoral thesis "Visual Body Pose Analysis for Human-Computer Interaction" and was awarded a Ph.D. degree from the ETH Zurich. He is currently working as a Post-Doc at ETH Zurich. [Google Scholar | contact]
 

Human-Computer Interaction

One important topic of his research is human-computer interaction, and more specifically hand gesture interaction. This involves real-time detection, tracking and recognition of hand gestures for interaction with interactive 3D models displayed on large screens (click below for videos).

Perceptive User Interface 3D Object Manipulation Interaction with a 3D City Model


Real-time Video Segmentation

A second important field of his research is real-time image and video segmentation. This work involves the design and development of highly efficient algorithms, such as the recently developed SEEDS superpixel algorithm.



3D Vision and Stereo Vision

Many of the problems mentioned above require segmentation, detection and recognition in a calibrated stereo system or even a 3D system. This can be achieved by using two or more color cameras, or by combining a color camera with a Time-of-Flight (ToF) or other IR-based sensor. Like this, 3D reconstructions or 3D hulls can be obtained which facilitate the detection of hand gestures or the recognition of full body posture.

Hand Gesture Recognition on Kinect Calibrated ToF Depth Full Body Pose Recognition



Publications

Van den Bergh M., Roig G., Boix X., Manen S. and Van Gool L. (2013). "Online Video Superpixels for Temporal Window Objectness". In International Conference on Computer Vision (ICCV). [PDF]

Van den Bergh M., Boix X., Roig G. and Van Gool L. (2013). "SEEDS: Superpixels Extracted via Energy-Driven Sampling". In arXiv. [PDF]

Van den Bergh M., Carton D. and Van Gool L. (2013). "Depth SEEDS: Recovering Incomplete Depth Data using Superpixels". In IEEE Workshop on Applications of Computer Vision (WACV). [PDF]

Van den Bergh M., Boix X., Roig G., de Capitani B. and Van Gool L. (2012). "SEEDS: Superpixels Extracted via Energy-Driven Sampling". In European Conference on Computer Vision (ECCV) (Vol. 7, pp. 13-26). [PDF]

De Luca A., Mattone R., Giordano P., Ulbrich H., Schwaiger M., Van den Bergh M., Koller-Meier E. and Van Gool L. (2012). "Motion Control of the CyberCarpet Platform". IEEE Transactions on Control Systems Technology, 20(9), 1-18. [PDF]

Van den Bergh M. and Van Gool L. (2012). "Real-time Stereo and Flow-based Video Segmentation with Superpixels". In IEEE Workshop on Applications of Computer Vision (WACV) (pp. 89-96). [PDF]

Uebersax D., Van den Bergh M., Gall J. and Van Gool L. (2011). "Real-time Sign Language Letter and Word Recognition from Depth Data". In IEEE Workshop on Human Computer-Interaction (pp. 383-390) [PDF]

Van den Bergh M., Carton D., De Nijs R., Mitsou N., Landsiedel C., Kuehnlenz K., Wollherr D., Van Gool L. and Buss M. (2011). "Real-time 3D Hand Gesture Interaction with a Robot for Understanding Directions from Humans". In IEEE International Symposium on Robot and Human Interactive Communication (Ro-Man) (357-362). [PDF]

Van den Bergh M. and Van Gool L. (2011). "Combining RGB and ToF Cameras for Real-time 3D Hand Gesture Interaction". In IEEE Workshop on Applications of Computer Vision (WACV) (pp. 66-72). [PDF]

Van den Bergh M. (2010). "Visual Body Pose Analysis for Human-Computer Interaction", Ph.D. Thesis, ETH Zurich, Switzerland. [PDF]

Van den Bergh M., Bosche F., Koller-Meier E. and Van Gool L. (2009). "Haarlet-based Hand Gesture Recognition for 3D Interaction". In IEEE Workshop on Applications of Computer Vision (WACV) (pp. 1-8). [PDF]

Van den Bergh M., Halatsch J., Kunze A., Bosché F., Van Gool L. and Schmitt G. (2009). "A Novel camera-based System for Collaborative Interaction with Multi-dimensional Data Models". In International Conference on Construction Applications of Virtual Reality (CONVR) (pp. 19-28). [PDF]

Van den Bergh M., Koller-Meier E. and Van Gool L. (2009). "Real-time Body Pose Recognition using 2D or 3D Haarlets". International Journal of Computer Vision (IJCV), 83(1), 72-84. [PDF]

Van den Bergh M., Kehl R., Koller-Meier E. and Van Gool L. (2009). "Real-time 3D Body Pose Estimation". Multi-Camera Networks: Concepts and Applications, Hamid Aghajan, Andrea Cavallaro, Ed., ELSEVIER, 335-360. [PDF]

Van den Bergh M., Koller-Meier E. and Van Gool L. (2008). "Fast Body Posture Estimation using Volumetric Features". In IEEE Workshop on Motion and Video Computing (WACV) (pp. 1-8). [PDF]

Van den Bergh M., Servaes W., Caenen G., De Roeck S. and Van Gool L. (2005). "Perceptive User Interface, a Generic Approach". In Computer Vision in Human-Computer Interaction (pp. 60-69). [PDF]
 

Awards

Annual Barco prize for Best Master Theses 2004, for the thesis titled "Perceptive User Interface, a Generic Approach", January 2005.

Best paper award for the paper titled "Depth SEEDS: Recovering Incomplete Depth Data using Superpixels", WACV, January 2013.

 

 

Projects

Interactive Urban Robot (IURO): The IURO project aims to build a robot that autonomously finds its way to pre-defined places, people or items in quickly changing environments through proactive communication with passers-by. In general, interactive robots provide people with information they need or want to have, like mobile museum guides. People address the robot and know what to obtain from it. For IURO we invert the perspective: The robot addresses arbitrarily passers-by in public (urban) areas in order to obtain 'vital' information from them: In which direction is square X? Where can I find shop Y? These are everyday knowledge gaps experienced by human pedestrians. We assume that mobile service robots will experience the same gaps while navigating outdoor in public spaces. (link).

CyberWalk: Walking is the most natural way for humans to navigate through their environment. To date, however, it is almost impossible to physically walk through Virtual Environments in an unconstrained way. In the EU project "CyberWalk" we developed an omni-directional treadmill, which together with markerless tracking, optimized control and several perceptual tricks enables humans to walk through Virtual Worlds in a natural and unconstrained fashion (link).

Value Lab: The Value Lab is an essential space in the new Information Science Laboratory (HIT) and represents the fusion between built architecture and digital design sciences. The Value Lab enables users to work together interactively. This lab will be integrated in challenging scientific projects within the ETH Zurich and with cooperation partners of the ETH Zurich. It will be used in teaching and for research purposes, such as analysis of large data sets, applications in marketing and monitoring, applications in real-time visualization, distributed real-time image rendering, and interactive screen design. The system will also support general usage for presentations, lectures, press and videoconferences. (link).

contact