Understanding and Modelling Human Attention for Soft Biometrics Purposes

Abstract

Soft biometrics systems have spread among recent years, both as a mean to empower classical biometrics, as well as stand-alone and complete solutions, with several application scopes ranging from digital signage to human-robot interaction. Among all, in the recent years emerged the possibility to consider as soft biometrics also the temporal evolution of both the human attention and the emotional states, and some recent works in the literature partially explored this exciting research line making use of either expensive tools or RGB-D devices. This work is instead the first attempt to perform soft biometrics identification of individuals on the basis of data acquired by a consumer camera, looking at users' attention evolution in time. The experimental evidence of the feasibility of using the proposed framework as soft-biometrics is given on a set of 22 users recorded by a tablet front-facing camera while watching, in an unconstrained mobile setting, a video running on the tablet screen.