Dr. Zdenek Kalal Email: kalal@tldvision.com Follow me:

Zdenek Kalal is a researcher in computer vision with the focus on real-time tracking of unknown objects. During his PhD and under suppervision of Dr. Krystian Mikolajczyk and Prof. Jiri Matas, Kalal developed an algorithm called TLD that stands for Tracking-Learning-Detection. The authors published 6 research papers, where TLD demonstrated significant improvement over state-of-the-art. TLD has been presented at competition UK ICT Pioneers 2011, where Kalal obtained a prize in category: "Technology Everywhere". Later on, TLD alborithm and corresonding video become well known on internet with more than 600.000 views on YouTube and a ~2000 registered members of correspoinding discussion group.

Kalal was invited for Google Tech Talk, Frontiers of Interaction conference, UK BBC radio and appeared in several magazines. The strong interest for TLD technology clearly showed that real-time and robust tracking of unknown object is needed in a large number of industrial applications. As of September 2011, Kalal is working as an independent researcher/developer and paticipates in academia as a reviewer.

01.10.2011 -- Founded a start-up company TLD Vision s.r.o.

07.04.2011 -- TLD source code and supporting wiki and discussion group (~2000 members).

24.03.2011 -- ICT Pioneers 2011

Read more about Predator.

Research

TLD TLD [2,3,4,5] is an algorithm that simultaneously tracks, learns and detects an unknown object in a video stream. TLD makes minimal assumptions about the object, the scene or the camera's motion. It requires only initialization by a bounding box and operates in real-time. See the project page for videos, download the source code, demo and the TLD data set.

Face detection We developed [1] a learning method that optimally combines boosting with bootstrapping and enables efficiently process of large training data sets. Using the method we built a real-time multi-view face detector with state of the art performance. Download the face detector code and test it your self.

Publications

• [6] Z. Kalal, K. Mikolajczyk, and J. Matas, “Tracking-Learning-Detection,” Pattern Analysis and Machine Intelligence, 2011.
  
• [5] Z. Kalal, K. Mikolajczyk, and J. Matas, “Face-TLD: Tracking-Learning-Detection Applied to Faces,” International Conference on Image Processing, 2010.
  [pdf][poster][ ]
@article{Kalal2010b,
author = {Kalal, Z. and Mikolajczyk, K. and Matas, J.},
journal = {International Conference on Image Processing},
title = {{Face-TLD: Tracking-Learning-Detection Applied to Faces}},
year = {2010} }
• [4] Z. Kalal, K. Mikolajczyk, and J. Matas, “Forward-Backward Error: Automatic Detection of Tracking Failures,” International Conference on Pattern Recognition, 2010, pp. 23-26.
  [pdf][ ]
@article{Kalal2010a,
author = {Kalal, Z. and Mikolajczyk, K. and Matas, J.},
journal = {International Conference on Pattern Recognition},
title = {{Forward-Backward Error: Automatic Detection of Tracking Failures}},
year = {2010}
}
• [3] Z. Kalal, J. Matas, and K. Mikolajczyk, “P-N Learning: Bootstrapping Binary Classifiers by Structural Constraints,” Conference on Computer Vision and Pattern Recognition, 2010.
  [pdf][poster 1][poster 2][ ]
@article{Kalal2010,
author = {Kalal, Z and Matas, J and Mikolajczyk, K},
journal = {Conference on Computer Vision and Pattern Recognition},
title = {{P-N Learning: Bootstrapping Binary Classifiers by Structural Constraints}},
year = {2010}
}
• [2] Z. Kalal, J. Matas, and K. Mikolajczyk, “Online learning of robust object detectors during unstable tracking,” On-line Learning for Computer Vision Workshop, 2009.
  [pdf] [ ]
@article{Kalal2009,
author = {Kalal, Z and Matas, J and Mikolajczyk, K},
journal = {On-line Learning for Computer Vision Workshop},
title = {{Online learning of robust object detectors during unstable tracking}},
year = {2009}
}
• [1] Z. Kalal, J. Matas, and K. Mikolajczyk, “Weighted Sampling for Large-Scale Boosting,” British Machine Vision Conference, 2008.
  [pdf][poster][code][ ]