Author/s: Noor Kamal Al-Qazzaza,∗, Amr aloula Ridhaa, Umniah Amjada
Affiliation: aDepartment of Biomedical Engineering, Al-Khwarizmi College of Engineering, University of Baghdad, Baghdad 47146, Iraq.
*Corresponding author’s email: noorbme@kecbu.uobaghdad.edu.iq
DOI: https://doi.org/10.31972/iceti2024.025
Abstract
Ageing is a difficult issue for facial recognition systems nowadays. Biological changes that occur with ageing provide a special problem since they can cause noticeable differences in face features between photos of the same individual taken at different ages. The extraction of robust facial characteristics for age-invariant face recognition is becoming increasingly important, especially when there are huge age differences between photographs of the same person’s face, because the face is the most affected area of the body by ageing. This study aims to create a deep learning algorithm that accurately predicts facial image-based age with low error, firstly, to build a robust and generalizable deep learning model that can properly identify gender from facial photos, secondly, and to advance age-related research and gender prediction, which may be used for demographic analysis, targeted advertising, personalized marketing, age-specific suggestions, and customized user experiences. Convolutional Neural Networks (CNN) was investigated to illustrate the effectiveness of deep learning-based methods on the UTKFace dataset. Additionally, the MobileNetV2 model consistently has the best mean accuracy rate when feature extraction is performed using the MobileNetV2 model, suggesting that it could be the most promising option for age-invariant face recognition.
Keywords: Age, gender, deep learning, CNN, classification
Full Paper