Child Face Recognition with Deep Learning

Abstract

Face recognition is a kind of identifying people in image. Face recognition technology play a role for children life. It is used for finding the missing children, school safety and school social network activities. Generally, older ages can recognize easily from each other but children are very hard to recognize. Convolutional Neural Networks (CNNs) is a top performer on face recognition. In this paper, the accuracy and performance of three Convolutional Neural Networks (CNNs) such as VGG Face based on two architectures (VGG16 and ResNet50), and MobileFaceNet on child face dataset is tested. The experiments results are shown and evaluated. According to experiments results, MobileFaceNet on child face dataset provide better accuracy than others. Among three proposed methods, the best recognition accuracy is 99.75% from MobileFaceNet.