Soft biometric attributes in the wild: case study on gender classification

Abstract

Soft biometrics has become an active field of research, as it provides useful attributes to assist in recognition systems. Its fusion with strong traits may serve to achieve reasonable recognition rates in less cooperative scenarios. These attributes can also be used to speed up database searches, or to describe an anonymous subject within a demographic group. Agreeing with recent research trends on the need to evaluate biometric systems using “in the wild” datasets, the current state-of-the-art in the emerging field of soft biometrics is presented, together with proposals and results on the particular problem of gender classification “in the wild”.