Site-adaptation methods for face recognition

• NCJ Number: 305612
• Author(s): Jilin Tu; Xiaoming Liu; Peter Tu
• Date Published: January 2010
• Length: 6 pages
• Grant Number(s): 2007-DE-BX-K191
• Agencies: NIJ-Sponsored
• Award: Site-Adaptive Face Recognition at a Distance

Annotation

The authors propose site adaptation for generic face recognizers based on facial images captured by cameras at the site as an adaptation set.

Abstract

Based on an OSFV[20] face recognizer with Gabor features selected by Adaboost algorithm, the authors propose several site adaptation methods at the feature level and at the model level. The experimental results showed that the proposed site adaptation approaches can significantly boost the performance of the authors’ generic face recognition algorithm at site with unforeseen illumination background and imaging conditions with a small adaptation set. (Publisher abstract provided)