Demographic Face Profiling Based on Age, Gender and Race

User profiling has lately got much interest and has been increasingly used in various fields of applications such as security, medicine, and commerce. The aim of this work is to predict a user demographic profile based on soft biometric modalities, namely the age, the gender and the race, for the authentication of suspicious people. We propose different types of characteristics based on global and local face features relative to the color, the texture and the shape. The retained characteristics are selected by the PSO algorithm. The classification phase is based on the SVM classifier optimized by a grid search to determine its best parameters. Validated on the public Morph II database and on our own database, the proposed approaches of users' demographic profile estimation yield interesting results.