Extraction of soft biometric characteristic in the wild.

Abstract

This thesis deals with soft biometrics. Soft biometrics refers to the physical, behavioral or adhered human characteristics like age, gender, ethnicity etc. These characteristics can be much easier to extract (especially from distance) than traditional biometric traits (like fingerprints and iris) yet they can be very useful in many applications. The thesis gives an extensive analysis of the state of the art in the field of soft biometrics by discussing the existing methods, the reported results and the public databases. Importantly, the thesis describes a number of novel contributions including new methods for age estimation from face images. Extensive experimental results are reported. Some open issues and future directions are also highlighted.