Comparative analysis of gender classification methods using convolutional neural networks

Abstract

Gender classification has become an important application in the fields of system automation and artificial intelligence, having important implications across various fields. The main challenge in this classification task is the variation in illumination that affects the quality of facial images. This study presents a method for identifying genders with Convolutional Neural Networks (CNNs). To address this issue, various preprocessing methods are applied, including Self Quotient Image (SQI), Histogram Equalization, Locally Tuned Inverse Sine Nonlinear (LTISN), Gamma Intensity Correction (GIC), and Difference of Gaussian (DoG), to stabilize the effects of illumination variations before the images are processed by the CNN. The CNN architecture used consists of 5 convolutional blocks and 2 fully connected blocks, which have proven effective in image recognition. The results of the study show that the model trained with the DoG method achieved an accuracy of 91.07%, making it the best preprocessing technique compared to other methods such as SQI and HE, which achieved accuracies of 90.39% and 88.76%, respectively. These findings demonstrate that the application of SQI in CNN can improve the accuracy of gender classification on facial images, providing better performance than previous methods. These findings are expected to serve as a foundation for further developments in facial image classification and its applications in various fields.