Auditory perception vs. face based systems for human age estimation in unsupervised environments: from countermeasure to multimodality

Abstract

• An emerging technique of age estimation using sound perception in unsupervised environment, is considered. • Sound perception can be efficiently be used as countermeasure to prevent attacks of face age estimation systems. • Sound perception can efficiently be integrated in a multimodal biometric system to improve the accuracy. Face-based age estimation systems are commonly considered in biometric applications as well as in other fields such as forensics or healthcare. For security purposes, features extracted from the face can be used to verify or estimate the age of individuals in order to control their access to physical or logical resources. The main problem in using facial biometrics is its sensitivity, to acquisition (e.g. illumination, pose, occlusion, image quality, etc.), to face expression, and especially to potential attacks in unsupervised environments. In this work, we propose a robust modality using both random auditory stimulation and Deep-learning based age estimation, though individual perception (RaS-DeeP): (1) as a countermeasure to prevent attacks on face-based age estimation systems, but also (2) : as a complementary modality in a multimodal biometric system (i.e. face-sound perception) in order to improve the performances of face-based age estimation system. Used as countermeasure, we show that RaS-DeeP provides promising results with an EER value of 4.2%. On the other hand, when considering the multimodal system face-auditory perception, we show that, the performance of face age estimation system is enhanced with an EER of 3.3%. To evaluate the performance of multimodal system in real-time, 71 subjects from different age ranges achieving five repetitions, participated in our experiment.