In this research, we present the implementation of various Face Attribute Classifiers (13 classifiers in total) to be able to recognize the presence/absence of particular traits, such as long/short hair, eyeglasses, hat, mustache, etc. Our approach consists of extracting CoHOG descriptors from different regions of the face according to each attribute, and training the system using Linear Support Vector Machine (SVM). The applicability of this research is demonstrated by a Real Time Face Attribute Recognition sample program.
Christian I. Penaloza, "Face Attribute Classifiers using Co-ocurrance Histogram of Gradients", Toshiba Research and Development Center, Kawaguchi, Japan. August, 2011.