Abstract:
A face-based system for age estimation was developed to determine whether the age of an individual presented in given face image is either an adult (ages 18 years & above) or an underage (ages below 18 years). It considers the age 18 as a criterion used by most institutions such as electioneering system, educational system, cigarette vending store, pharmaceutical store, border crossing, visa applications and many more.
It considers certain features extracted by an individual’s face image either through a file upload or a camera capture to determine if the person presented in the face image is either an adult or an underage. This is done following an execution of three major stages after the input. First, it detects the face region from the entire image using Viola-Jones algorithm. Secondly, it carries out feature extraction from the detected region using a multi-feature technique (that is, a serial combination of Gabor algorithm for wrinkle feature and then the Local Binary Pattern). Finally, system training and age classification is done using a Support Vector Machine.
The use of a multi-feature technique, that is, more than one feature extraction technique is the research gap. The result of the experiment carried out proved the research to be a success and contribution to knowledge. The experimental test carried out using 10 face images of both underage and adult gives a minimum of 85% accuracy and 15% misclassification rate
