Face Gender Recognition Using Multi Layer Perceptron with OTSU Segmentation

Abstract

Face Gender recognition has been a swiftly enhancing and interesting area which is of high challenge and significance in real-time applications of image processing. There are lot of open opportunities in person identification such as posture, clothing, hair, voice and gait; but none are as compelling as face recognition. Face gender recognition is significantly an efficient cognitive process and there is definitely a need of robust methods for efficient categorization of male and female subjects. In the paper here, OTSU segmentation has been applied for the feature extraction on subset of 780 images from Faces 94 dataset and Multi Layer Perceptron (MLP) network is further employed to investigate the local minima and global maxima of Face Gender Recognition accuracy with multiple considered hidden layers. The experimentation has resulted in getting higher face gender recognition accuracy as 99.658% with fifteen hidden layers of MLP.