Learning-based face synthesis for pose-robust recognition from single image

Asthana, Akshay, Gedeon, Tom, Goecke, Roland and Sanderson, Conrad (2009). Learning-based face synthesis for pose-robust recognition from single image. In: Proceedings of the British Machine Vision Conference 2009. British Machine Vision Conference 2009, Trinity College, Dublin, UK, (1-10). 2-4 September, 2009.

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
Paper324.pdf Paper324.pdf application/pdf 982.54KB 156

Author Asthana, Akshay
Gedeon, Tom
Goecke, Roland
Sanderson, Conrad
Title of paper Learning-based face synthesis for pose-robust recognition from single image
Conference name British Machine Vision Conference 2009
Conference location Trinity College, Dublin, UK
Conference dates 2-4 September, 2009
Proceedings title Proceedings of the British Machine Vision Conference 2009
Place of Publication London, U.K.
Publisher British Machine Vision Association and Society for Pattern Recognition
Publication Year 2009
Year available 2009
Sub-type Fully published paper
Start page 1
End page 10
Total pages 10
Language eng
Abstract/Summary Face recognition in real-world conditions requires the ability to deal with a number of conditions, such as variations in pose, illumination and expression. In this paper, we focus on variations in head pose and use a computationally efficient regression-based approach for synthesising face images in different poses, which are used to extend the face recognition training set. In this data-driven approach, the correspondences between facial landmark points in frontal and non-frontal views are learnt offline from manually annotated training data via Gaussian Process Regression. We then use this learner to synthesise non-frontal face images from any unseen frontal image. To demonstrate the utility of this approach, two frontal face recognition systems (the commonly used PCA and the recent Multi-Region Histograms) are augmented with synthesised non-frontal views for each person. This synthesis and augmentation approach is experimentally validated on the FERET dataset, showing a considerable improvement in recognition rates for ±40◦ and ±60◦ views, while maintaining high recognition rates for ±15◦ and ±25◦ views.
Subjects 080109 Pattern Recognition and Data Mining
080104 Computer Vision
080106 Image Processing
810107 National Security
970108 Expanding Knowledge in the Information and Computing Sciences
Keyword Gaussian Process Regression
Active Appearance Models
Q-Index Code E1
Q-Index Status Provisional Code
Institutional Status UQ

Version Filter Type
Citation counts: Google Scholar Search Google Scholar
Created: Tue, 24 Nov 2009, 18:37:45 EST by Dr Ildiko Horvath on behalf of School of Information Technol and Elec Engineering