A surveillance system using CNN for face recognition with object, human and face detection

Abstract

Recently, surveillance system plays an important role in solving several crimes by replacing human to watch monitors. Not only many functions are complicatedly integrated to a system but also a system is evolved to capture statistical data to extract useful information. But integrating many functions should be considered to make it have reduced processing time because a system has limited processing ability. If a system considers moving objects, it could reduce processing time because surveillance system normally has static background that is useless information. People and face detection are performed in detected objects. Detected faces are recognized using CNN(Convolutional Neural Network). The processing time of the proposed system is reduced and true rate of face recognition is 72.7% under varying distance from 2m to 5m.