Stepan Obdrzalek

Qualifications:
2013-present		Traveling and volunteering through Asia and Australia
2010-2012		Research fellow at the Tele-immersion lab, Dept. of Electrical Engineering and Computer Science, University of California, Berkeley, USA Field of interest: Human pose estimation, computer-assisted coaching of elderly
2004-2010		Research fellow at the Center for Machine Perception, Faculty of Electrical Engineering, Czech Technical University in Prague Research in cooperation with Toyota Motors Europe Field of interest: Intelligent vehicles, object recognition, motion detection and segmentation, egomotion estimation
2000-2004		Full time Ph.D. student at the Center of Machine Perception, Faculty of Electrical Engineering, Czech Technical University in Prague Field of study: Object recognition, computer vision PhD Thesis: Object Recognition Using Local Affine Frames
1993-2000		Master degree at Department of Computer Science, Faculty of Electrical Engineering, Czech Technical University in Prague Field of study: Computer science, computer graphics and computer vision Diploma thesis: Compensation of illumination effects on facial appearance using a 3D model

Awards:
		Česká hlava 2006 doctorandus: national prize for best doctorate of 2006
		Best Science Paper Award on British Machine Vision Conference 2005, for paper Štěpán Obdržálek and Jiří Matas: Sub-linear Indexing for Large Scale Object Recognition
		Participation in team winnig 2nd place in ICCV05 Computer Vision Contest ”Where Am I?”

Selected publications:
2005		Štěpán Obdržálek and Jiří Matas: Sub-linear Indexing for Large Scale Object Recognition Best science paper award on British Machine Vision Conference 2005
2006		Štěpán Obdržálek and Jiří Matas: Object Recognition using Local Affine Frames on Maximally Stable Extremal Regions, invited chapter in Jean Ponce et al., editors: Toward Category-Level Object Recognition
2007		Štěpán Obdržálek: Object Recognition Using Local Affine Frames PhD thesis, Czech Technical University
2008		Štěpán Obdržálek, Michal Perd'och and Jiří Matas: Dense Linear Time Correspondences for Tracking Workshop on Visual Localization for Mobile Platforms, CVPR 2008, Anchorage, Alaska (A correspondence search method based on hashing of low-dimensional local image descriptors. Constant time and memory complexity for a fixed image size. Real time CPU implementation, highly paralellisable for SIMDs (GPU)) "I found this a fascinating paper, presenting a very original take on the familiar matching problem." --- Anonymous reviewer
2010		Štěpán Obdržálek and Jiří Matas: A Voting Strategy for Visual Ego-Motion from Stereo IEEE Intelligent Vehicles Symposium, San Diego, California, June 2010 (Hough transform based estimation of stereo camera motion. Explicitly considers triangulation uncertainty and precission decreasing with distance)
2012		Štěpán Obdržálek, Gregorij Kurillo, Ferda Ofli, Ruzena Bajcsy, Edmund Seto, Holly Jimison and Michael Pavel: Accuracy and Robustness of Kinect Pose Estimation in the Context of Coaching of Elderly Population EMBC, 34th International Conference of the IEEE Engineering in Medicine and Biology Society, San Diego, California, August 2012 (Qualitative comparison of human poses (skeletons) from Kinect and from PhaseSpace motion capture system that uses active LED markers) "Excellent work of high clinical potential." --- Anonymous reviewer
remaining publications can be found at http://cmp.felk.cvut.cz/biblio/publicat/, entering 'obdrzalek' into the 'Author' field

Research fellowships:
July – September 1999, February – March 2000		Image Science group, Department of Electrotechnics, Swiss Federal Institute of Technology (Eidgenossische Technische Hochschule - ETH), Zürich, Switzerland
January – July 2002		Centre for Vision, Speech, and Signal Processing, School of Electronics and Physical Sciences, University of Surrey, Guildford, United Kingdom In Erasmus Student Exchange Program

Personal:
		Photogallery

Štěpán Obdržálek