Article
DeepUAge: Improving Underage Age Estimation Accuracy to Aid CSEM Investigation
Contribution Summary
This paper presents DeepUAge, a deep learning model for improving underage age estimation accuracy to aid Child Sexual Exploitation Material (CSEM) investigation. The model is trained on the VisAGe dataset, the largest underage facial age dataset, and achieves state-of-the-art performance for age estimation of minors. The mean absolute error (MAE) rate of 2.73 years is significantly lower than existing approaches. The work tackles the challenges of collecting and annotating underage facial age data, which is a critical issue in CSEM investigation. The proposed model can expedite digital investigations by accurately classifying explicit content possession as illegal. The authors also evaluate several facial image pre-processing techniques and the impact of facial landmark points on age estimations. The results demonstrate the effectiveness of DeepUAge in improving underage age estimation accuracy, making it a valuable tool for law enforcement agencies.
Keywords: Child Sexual Exploitation Material (CSEM); Underage facial age estimation; Deep learning; VisAGe dataset; Age estimation accuracy; Digital investigations; Facial image pre-processing
Abstract
Age is a soft biometric trait that can aid law enforcement in the identification of victims of Child Sexual Exploitation Material (CSEM) creation/distribution. Accurate age estimation of subjects can classify explicit content possession as illegal during an investigation. Automation of this age classification has the potential to expedite content discovery and focus the investigation of digital evidence through the prioritisation of evidence containing CSEM. 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 recent years. These existing approaches perform satisfactorily for adult subjects, but 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, DeepUAge, that achieved state-of-the-art beating performance for age estimation of minors. This paper describes the design and implementation of this model. 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
@article{anda2020UnderageAgeEstimation,
author={Anda, Felix and Le-Khac, Nhien-An and Scanlon, Mark},
title="{DeepUAge: Improving Underage Age Estimation Accuracy to Aid CSEM Investigation}",
journal="{Forensic Science International: Digital Investigation}",
year="2020",
month="04",
volume = "32",
pages = "300921",
issn = "2666-2817",
doi = "https://doi.org/10.1016/j.fsidi.2020.300921",
url = "http://www.sciencedirect.com/science/article/pii/S2666281720300160",
publisher={Elsevier},
keywords = "Child Sexual Exploitation Material (CSEM), Age estimation, Underage facial age dataset, Child sexual abuse investigations, Deep learning",
abstract={Age is a soft biometric trait that can aid law enforcement in the identification of victims of Child Sexual Exploitation Material (CSEM) creation/distribution. Accurate age estimation of subjects can classify explicit content possession as illegal during an investigation. Automation of this age classification has the potential to expedite content discovery and focus the investigation of digital evidence through the prioritisation of evidence containing CSEM. 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 recent years. These existing approaches perform satisfactorily for adult subjects, but 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, DeepUAge, that achieved state-of-the-art beating performance for age estimation of minors. This paper describes the design and implementation of this model. 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.}
}