Special Tutorial - State of the Art Face Detection: Cascade Boosting Approaches

摘要

The face detection problem consists on finding the position and size of all faces (if any) appearing in an image or video. It is a complex problem, mainly because of the real-time and real-world (variability of human faces, image acquisition conditions and image background) requirements. In the present tutorial we first introduce the face detection problem and we explain its peculiarities. Then we present the most commonly employed approaches, especially the ones based on statistical learning. Finally, we describe the so-called cascade boosting approaches, which have recently solved the face detection problem in a fast and robust manner.