Abstract

This paper aims to measure the accuracy of a face analysis system, implemented in Processing using deep neural networks. The face analysis targets age and gender identification. If proven efficient, such a system can be used in on-line security applications. The tests are implemented using an on-line image dataset, containing hundreds of different faces in over 4000 different poses. The resulting accuracy of the system is expounded as a function of age. For age identification, the accuracy of the system was 48% across all age groups, while for gender identification the measured accuracy was 80%.