AGE AND GENDER CLASSIFICATION FROM IRIS IMAGES OF THE EYE USING MACHINE LEARNING TECHNIQUES

Abstract

The human face stands out as a preferable choice for biometric human authentication, the iris is a secure biometric that contains features with a low forgery rate and is unique to individuals. This paper presents a model for age and gender classification from iris images captured from male and female genders between the ages of 5 and 60 years. The age and gender were grouped as FemaleYoung, FemaleTeen, FemaleAdult, MaleYoung, MaleTeen, and MaleAdult. On the six categories, feature extraction was done using a 3D histogram with Principal Component Analysis (PCA) applied for dimension reduction to further reduce duplicated or unwanted features. Next, the Support Vector Machine (SVM) was applied to classify the iris images into the six groups recording various performance values. The 3D histogram with PCA recorded an excellent classification performance accuracy of 99.27% as against the EfficientNet deep learning model which recorded 52.29%. The recommended feature technique can help to adequately classify gender and age from iris images leading to a more robust recognition model.