Component-based Online Learning and Its Application to Face Detection

K.-M. Lee (Korea)

Keywords

Componentbased detection, online learning, unsupervised learning, face detection, color image

Abstract

This paper proposes component-based online learning, which is based on using unsupervised learning to find a set of templates specific to objects consisting of components and their relations. The main difference from previously reported methods is the use of on-line learning, which is ideal for highly repetitive tasks. This results in faster and more accurate object detection, because system performance improves with continued use. Further, uncertainty is added by calculating the standard deviation of components and relations. A component-based method with uncertainty provides flexibility to allow variability to describe an object in appearance and geometry. The effectiveness of the proposed learning algorithm is demonstrated with respect to face detection in color images.

Important Links:



Go Back