Prediction of Gender and Age Period from Periorbital Region with VGG16

Abstract

Using deep learning methods, age and gender estimation from people’s facial area has become popular. Recently, with the increase in the use of masks due to Covid-19, only the eye area of people is seen. The periorbital region can give an idea about the person’s characteristics, such as age and gender. This study it is aimed to predict gender and age from images obtained by cutting the eye area from facial photographs of people using Visual Geometry Group-16 (VGG16). With the transfer learning method for age group (male, female) and gender group (child, youth, adults, and old) classification, 5714 images in the data set were used for the age group, and 3280 images were used for the gender group. As a result of this study, 99.41% success in age estimation and 95.73% in gender estimation was achieved.