Face Recognition in the Fog Cluster Computing

Abstract

This work proposes a cluster computing based facial recognition system in a fog computing environment under Internet of Things theme. The proposed system uses Haar Cascades to detect faces in a test image and uses Local Binary Pattern Histogram algorithm to recognize them in parallel by making use of computing power of several Raspberry Pi nodes in the cluster. The cluster uses Simple Linux Utility for Resource Management scheduler to schedule various jobs among the nodes and Message Passing Interface to facilitate communication among various processes running in the cluster. The system uses aggregated computation of the cluster. The facial recognition tasks were executed parallelly for obtaining better performance in terms of execution time. The proposed system can be extended to detect and recognize multiple objects in an image, which has several benefits in other applications.