Dynamic Scalable Distributed Face Recognition System Security Framework

Konrad Rzeszutek
University of New Orleans
Computer Science Department
2000 Lakeshore Drive
New Orleans, LA 70148

Abstract:

Many of our daily interactions depend on being able to recognize one's face. Modeling human recognition can be used in many fields: surveillance systems, security systems, autonomous navigation of vehicles and many more.

Current face recognition technologies are extremely computationally intensive. To accommodate the load in near-real time, a multi component scalable distributed framework is presented. This framework allows separating the functionality of recognition, notification, and replay of camera-feed in different independent components. Furthermore the framework was designed with RAS (reliability, availability and serviceability), adaptability to network changes and easy path of adding (or replacing) of components in mind. Also becuase of its distributed nature, it can work over a large network (Internet).

This paper describes in detail the framework of such system.