Representing the Appearances of Human Faces for Recognition (00:16)
Face recognition has been heavily studied in computer vision over the past decade. The difficulties of face recognition lie in the inherent variation of facial characteristics and the environment of image acquisition. Traditional face recognition systems usually impose restrictions on the imaging situation, such as well-controlled environment or near-frontal, well-aligned images. However, for most of the face images taken from a general digital camera in our daily life, these assumptions are not always true. In this talk, we are going to present a general framework that tries to solve these problems practically.