Under-age Facial Age Estimation: Improving Deep Learning Performance to Aid CSEM Investigations

Authors: Anda, Felix; Le-Khac, Nhien-An and Scanlon, Mark

Publication Date: March 2020

Publication Name: Forensic Science International: Digital Investigation

Abstract:

Age is a soft biometric trait that can help identify victims of Child Sexual Exploitation Material (CSEM). Accurate age estimation of subjects can classify explicit content as illegal possession of this material during an investigation. Automation of this age classification has the potential to expedite content discovery and prioritise evidence processing in CSEM cases. In recent years, artificial intelligence based approaches for automated age estimation have been created, and many public cloud service providers offer this service on their platforms. The accuracy of these algorithms have been improving over the past years, but nonetheless, these existing approaches perform wholly inadequately for underage subjects. To this end, the largest underage facial age dataset, VisAGe, has been used in this work to train a ResNet50 based deep learning model DCA16K that achieved a mean absolute error (MAE) of 1.528, outperforming existing state-of-the-art estimation models. This paper describes the design and implementation of this model. Finally, an evaluation, validation and comparison of the proposed model is performed against existing facial age classifiers resulting in the best overall performance for underage subjects.

BibTeX Entry:

@article{anda2020UnderageAgeEstimation,
author={Anda, Felix and Le-Khac, Nhien-An and Scanlon, Mark},
title="{Under-age Facial Age Estimation: Improving Deep Learning Performance to Aid CSEM Investigations}",
journal="{Forensic Science International: Digital Investigation}",
year="2020",
month="03",
publisher={Elsevier},
abstract={Age is a soft biometric trait that can help identify victims of Child Sexual Exploitation Material (CSEM). Accurate age estimation of subjects can classify explicit content as illegal possession of this material during an investigation. Automation of this age classification has the potential to expedite content discovery and prioritise evidence processing in CSEM cases. In recent years, artificial intelligence based approaches for automated age estimation have been created, and many public cloud service providers offer this service on their platforms. The accuracy of these algorithms have been improving over the past years, but nonetheless, these existing approaches perform wholly inadequately for underage subjects. To this end, the largest underage facial age dataset, VisAGe, has been used in this work to train a ResNet50 based deep learning model DCA16K that achieved a mean absolute error (MAE) of 1.528, outperforming existing state-of-the-art estimation models. This paper describes the design and implementation of this model. Finally, an evaluation, validation and comparison of the proposed model is performed against existing facial age classifiers resulting in the best overall performance for underage subjects.}
}