ExpressionAuth: Utilizing Avatar Expression Blendshapes for Behavioral Biometrics in VR

As interests in Virtual Reality (VR) continue to rise, head-mounted displays (HMDs) are actively being developed. Current user authentication method in HMDs requires the use of virtual keyboard, which has low usability and is prone to shoulder surfing attacks. This paper introduces ExpressionAuth, a novel authentication method which uses face tracking capabilities in certain HMDs to verify the identity of the user. ExpressionAuth leverages smile as the expression for user verification. ExpressionAuth has the potential to be a secure and usable biometrics, achieving EER as low as 0.00178 and AUC of up to 0.999 in our experiments.