Hybrid Age Estimation Using Facial Images

Age estimation determines a person's age or age group using facial images and has many real-world applications. This paper investigates various algorithms used to improve age estimation. A combination of features and classifiers are compared. A database of facial images is trained to extract features using algorithms such as local binary patterns (LBP), active shape models and histogram of oriented gradients (HOG). The age estimation is done using three age groups: child, adult, senior. The ages are classified using support vector machine (SVM), K-nearest neighbour (KNN), gradient boosting tree (GBT). The age estimation model is evaluated using the FG-NET aging database obtaining positive results of 82% success rate.