Who Is in the Crowd? Deep Face Analysis for Crowd Understanding

Crowd understanding plays a vital role in management processes and continues to receive a growing interest in various public and commercial service sectors. Surveillance, entertainment, marketing, and social sciences are only a few of the fields that can benefit from the development of automatic systems for crowd understanding. Little attention has been paid to the study of who are the subjects that characterize the crowd. In this article, we present a crowd understanding system based on face analysis capable of providing statistics about people in the crowd in terms of demographic (i.e. gender and age group), affective state (i.e. eight emotions, valence and arousal), and fine-grained facial attributes.