Gender Perception From Faces Using Boosted LBPH (Local Binary Patten Histograms)

Automatic Gender classification from faces has several applications such as surveillance, human computer interaction, targeted advertisement etc. Humans can recognize gender from faces quite accurately but for computer vision it is a difficult task. Many studies have targeted this problem but most of these studies used images of faces taken under constrained conditions. Real-world applications however require to process images from real-world, that have significant variation in lighting and pose, which makes the gender classification task very difficult. We have examined the problem of automatic gender classification from faces on real-world images. Using a face detector faces from images are extracted aligned and represented using Local binary pattern histogram. Discriminative features are selected using Adaboost and the boosted LBP features are used to train a support vector machine that provides a recognition rate of 93.29%