David Hurych
Ph.D. student

Telephone:   +420 22435 7679
Address: Karlovo namesti 13
121-35 Praha 2
Czech Republic
Office: E119

2007 - present:Ph.D.Student at the Czech Technical University CTU-FEL
(specialization: biocybernetics and artificial intelligence.)
2002 - 2007:Graduate at the Banska Technical University - Ostrava
(FEI, computer science)

X33KUI - Cybernetics and artificial intelligence (Labs) - link
X383ZS - Signal and Image Processing (Labs) - link
A4M33DZO - Digitální Obraz (Labs) - link
A6M33ZMO - Zpracování Medicínských Obrazu (Labs) - link

Real-time object tracking
Visual object detection

  • David Hurych, Tomas Svoboda, Jana Trojanova, Yadhunandan US:
    Active shape model and linear predictors for face association refinement
    International Conference on Computer Vision, Kyoto, Japan, September 29 - October 2, 2009. pdf
  • David Hurych:
    Thesis Proposal: Incremental Learning of Linear Predictors for Fast Object Tracking
    CTU in Prague, 2009. pdf
  • David Hurych, Tomas Svoboda:
    Incremental learning and validation of sequential predictors in video browsing application
    International Conference on Computer Vision Theory and Application, Angers, France, 2010. pdf
  • David Hurych, Karel Zimmermann, Tomas Svoboda:
    Detection of Unseen Patches Trackable by Linear Predictors
    Computer Vision Winter Workshop, Mitterberg, Styria, Austria, 2011. pdf
  • David Hurych, Karel Zimmermann, Tomas Svoboda:
    Fast Learnable Object Tracking and Detection in High-Resolution Omnidirectional Images
    International Conference on Computer Vision Theory and Application, Vilamoura, Algarve, Portugal, 2011. pdf
    received Best Student Paper Award
  • Karel Zimmermann, David Hurych, Tomas Svoboda:
    Improving Cascade of Classifiers by Sliding Window Alignment in Between
    International Conference on Automation, Robotics and Applications, Wellington, New Zealand, 2011. pdf
  • Karel Zimmermann, David Hurych, Tomas Svoboda:
    Exploiting Features -- Locally Interleaved Sequential Alignment for Object Detection
    Asian Conference on Computer Vision, Daejeon, South Korea, 2012. pdf
  • Karel Zimmermann, David Hurych, Tomas Svoboda:
    Non-Rigid Object Detection with Local Interleaved Sequential Alignment (LISA)
    Transactions on Pattern Analysis and Machine Intelligence,, 2014 (accepted). pdf

Take a look at some results of our tracking method. When tracking, we estimate the 2D homography. From 4-point correspondences we may compute the full perspective transformation. This gives us the knowledge of the object's position in 3D. The tracker is incrementally updated (learned) for new object appearances. The original tracker was trained on a single object image displayed in the top right corner in the videos. The application runs in real-time. The program was running on a notebook with 2.66 GHz dual-core and images were captured by a low cost 15 fps webcam. The framerate displayed in the videos shows the speed of the tracking part of the algorithm. The program was implemented in Matlab.

01 - Object tracking

02 - Object mapping (augmented reality)

03 - 3D face tracking


See some publications and demos at our
Advanced Linear Predictors project page

Center for Machine Perception
Faculty of Electrical Engineering
Czech Technical Univerzity in Prague