Design of Home Security System Using Face Recognition with Convolutional Neural Network Method

Abstract

Home security system with good accuracy and efficiency in controlling access to the door system is needed in order to identify people who enter the house accurately. Home security conventionally uses a key to open the door, making security low due to several factors. Various face recognition methods has been studied to determine the most accurate method in identifying people who has access to the house. In this study, Haar Cascade and CNN (Convolutional Neural Network) method were applied to face detection and classify 5 class of family member that can access the house. Based on the results of the analysis, the CNN model in this study uses an 64x64 sizes of input, 0.001 learning rate value, 3x3 filter size, 10 number of epochs, 1200 training data with 240 data for each class, and 150 testing data with 30 data for each class. The classification process yields the accuracy of 99% in identifying the family member of the house, hence giving access to open the door.