Intelligent Home Surveillance System using Convolution Neural Network Algorithms

Abstract

The creation of an automated security system aims to protect residences and workplaces by automating visitor entrance and enabling more flexibility in visitor record maintenance. Among all biometric authentications, face recognition is very secure because of unique facial features. There are two phases in authentication, face mask detection and face recognition. In first phase, Grassmann algorithm is used for face mask detection. If any mask is discovered, an alarm will sound for the user to remove the mask and in second phase face recognition is done through CNN. The CNN method is utilized to compare facial traits, and if an outsider is found, a warning message is then displayed to the user. Real time datasets are collected for training and testing the CNN model. The executed result gives 98.02% higher accuracy compared to existing method.