Continuous Cloud User Authentication By Efficient Facial Recognition

Abstract

Designing any cloud-based application typically includes a vital step for securing user identification and access control. A potential approach to increase security against mali-cious and unauthorized access to protected applications is multi-factor authentication based on biometrics. A versatile and reliable method of biometric identification is facial recognition since it can be implemented with less specialized hardware. Facial recognition-based continuous authentication is a promising strat-egy for assuring security. Because Convolutional Neural Networks are capable of comprehending the high level characteristics required to recognize human faces, they are known to be the most accurate and reliable approach for used identification. Their broad use is restricted by the computational requirements that results from utilizing this method for facial recognition. In order to minimize the computational load and boost the functional speed of authentication, this research suggests creating a modular facial recognition system that integrates techniques like frame differencing and face detection and face recognition process. A system of modular processing phases is designed to make the continuous authentication process efficient, as respective modules are activated when triggered by changes in the input data.