Rotated Face Detection Using AdaBoost

Face detection is a challenging problem in the field of computer vision and a key component of machine intelligence. Viola and Jones put forward a face detection framework in 2001 that can process images extremely rapidly while achieving high detection rates. This paper extends the framework to handle rotated faces through a simple improvement. The method is to rotate the detected sub-windows by 30 degree clockwise and counterclockwise within the image plane. Experimental results show that the proposed algorithm has a good detection rate and reduces the false positives regardless of the rotation of the faces.