Multi-View Face Detection: A Comprehensive Survey

Locating multi-view faces in images with a complex background remains a challenging problem. In recent years many techniques have been applied to this problem, such as view-based learning, neural networks and AdaBoost. Most of the techniques from statistical analysis of face and non-face samples have good performance in detection of frontal faces but locating multi-view faces detection still remains challenging. The task of this paper is to present a comprehensive and critical survey of multi-view face detection. One of the major challenges encountered by face detection lies in the difficulty of handling arbitrary poses variations. As shown in Fig.1, in real-world images, faces have significant variations in orientation, pose, facial expression, lighting conditions, etc. This paper describes various methods and algorithms to detect faces with non-upright (rotated) and non-frontal (profile) faces and survey the progress toward a system which can detect faces regardless of pose reliably. As the number of proposed technique increased, the survey and evaluation become important.