Event Location: NSH 1507
Bio: Sergey Tulyakov is a PhD student advised by Prof. Nicu Sebe at the Department of Information and Communication Technologies, University of Trento, Italy. His research interest includes 2D and 3D computer vision with particular emphasis on real-time analysis of human faces. He received his MS and BS in computer science from the Belarusian State University of Informatics and Radioelectronics.

Abstract: Most of the facial expression recognition methods assume frontal or near-frontal head poses and usually their accuracy strongly decreases when applied to non-frontal poses. In this work, we propose a 3D approach for head pose tracking and a new framework for recognizing facial expression under extreme head poses. By augmenting the precision of a template-based tracker with the ability to recover offered by a frame-by-frame head pose estimator, we are able to address pose ranges for which face features are no longer visible, while maintaining state-of-the-art performance. By projecting a face onto a pose-invariant representation, we transform the facial expression recognition problem under large head poses into a missing data classification problem. Once projected, the visible part of the face is split into overlapping patches, which are input to independent local classifiers, and a voting scheme gives the final output. Experimental results on common benchmarks show that our method can accurately recognize facial expressions in a much larger pan and tilt range than state-of-the-art approaches, obtaining comparable performance to the best existing systems working only in narrower ranges.