Facial Appearance Analysis for Age Group Prediction Using Convolutional Neural Network

Facial analysis is the process of using computer algorithms to analyze human faces in real time in images or videos. For age prediction, most of the works in the domain are regression-based analysis. Therefore, the aim of this study is to design a convolutional neural network (CNN) model for age group prediction using facial appearance analysis. A total of 13,363 facial images were retrieved from an online repository and direct capturing, comprising of individuals from different age groups, races, and gender. The results from the study revealed that the edge of the face between the left eyeball, the tip of the mouth and the right eyeball predicted the age of a human being. The CNN model adaptively learns from the spatial hierarchies of features, from low to high level patterns, and adequately classified and predicted the age group with an accuracy score of 88.6%, which is an improvement over the benchmark of 86%. The model, having suitably addressed the classification problems of the classical methodologies, has made it well-suited for object detection, image segmentation, and classification. Therefore, the model is recommended for deployment in biometric security and specialized applications for person’s identification, age prediction and specialized services.