Enhanced Privacy-enabled Face Recognition using κ-Identity Optimization

Authors

Ryan Karl, University of Nebraska-LincolnFollow

First Advisor

Benjamin Riggan

Date of this Version

12-2023

Document Type

Article

Citation

A thesis presented to the faculty of the Graduate College at the University of Nebraska in partial fulfillment of requirements for the degree of Master of Science

Major: Electrical Engineering

Under the supervision of Professor Benjamin Riggan

Lincoln, Nebraska, December 2023

Comments

Copyright 2023, Ryan Karl

Abstract

Facial recognition is becoming more and more prevalent in the daily lives of the common person. Law enforcement utilizes facial recognition to find and track suspects. The newest smartphones have the ability to unlock using the user's face. Some door locks utilize facial recognition to allow correct users to enter restricted spaces. The list of applications that use facial recognition will only increase as hardware becomes more cost-effective and more computationally powerful. As this technology becomes more prevalent in our lives, it is important to understand and protect the data provided to these companies. Any data transmitted should be encrypted and privatized to prevent recovery of any facial data provided to these systems. This thesis proposes a novel method of recognition in a privatized way that combines multiple κ different identities into one combined template. This combined template can be used for recognition with a small accuracy impact but reduces the amount of individual facial data transmitted. To further improve the accuracy and privacy, a novel training strategy that utilizes the κ-identity templates is proposed. Finally, to determine how much facial data is in a template, a new metric is introduced. The goal of this metric is to find how much facial data is contained in a template for a given facial recognition system, with a private system's goal of reducing the amount of facial data in the system.

Advisor: Benjamin Riggan