Abstract:
A real-time algorithm for accurate localization of facial landmarks in a single monocular image will be presented. The algorithm is formulated as an optimization problem where the sum of responses of local classifiers is maximized with respect to the camera pose by fitting an articulated 3D shape model. The algorithm simultaneously estimates the head position and orientation and detects the facial landmarks in the image. Self-occluded landmarks are not considered in the maximization which allows the operation for a large set of viewing angles. Despite being local, we show that the basin of attraction is sufficiently large to be initialized by a scanning window face detector. Experiments demonstrate promising performance on standard datasets. The algorithm was tested against a publicly available binary of recent algorithm “Chehra” by Antonakos et al., CVPR 2014, with slightly better results on both AFLW and 300-W datasets.
Bio:
Jirı´ Matas received the MSc degree (with honors) in cybernetics from the Czech Technical University, Prague, in 1987, and the PhD degree from the University of Surrey, United Kingdom, in 1995. From 1991 to 1997, he was a research fellow at the Centre for Vision, Speech and Signal Processing, University of Surrey, working with J. Kittler. In 1997, he joined the Center for Machine Perception, Czech Technical University. Since 1997,
he has held various positions at these two institutions. His research interests include object recognition, sequential pattern recognition, invariant feature detection, and Hough Transform and RANSAC-type optimization. He has published more than 100 papers in refereed journals and conference proceedings. His publications have more than 800 citations in the science citation index. He is on the editorial board of the IEEE Transactions on Pattern Analysis and Machine Intelligence and the International Journal of Computer Vision. He has served in various roles at major international conferences (for example, ICCV, CVPR, ICPR,
and NIPS) and cochaired ECCV 2004 and CVPR 2007. He received the Best Paper Prize at the British Machine Vision Conferences in 2002 and 2005.