Over twenty years of eigenfaces

Abstract

The inaugural ACM Multimedia Conference coincided with a surge of interest in computer vision technologies for detecting and recognizing people and their activities in images and video. Face recognition was the first of these topics to broadly engage the vision and multimedia research communities. The Eigenfaces approach was, deservedly or not, the method that captured much of the initial attention, and it continues to be taught and used as a benchmark over 20 years later. This article is a brief personal view of the genesis of Eigenfaces for face recognition and its relevance to the multimedia community.