Juvenile detection by LBP and SVM

At present, it is very urgent to protect the juniors from the negative influence of the internet. This paper aims at the automated Juvenile detection based on facial appearance as the gateway for the access of adult websites. To describe the facial appearance, local binary patterns (LBP) are applied to extract the local texture from three main facial regions including forehead, eye corner and cheek. Then we used support vector machine (SVM) to act as the classifier to identify one person is a juvenile or an adults. Experiments show that our solution is effective. The recognition rate is about 87% on the FG-NET database.