Face Verification Bypass

TL;DR

This paper demonstrates that generative models can produce diverse images with similar feature vectors to a target, effectively bypassing face verification systems in both controlled and real-world scenarios.

Contribution

It introduces a method using StarGAN v2 to generate images that mimic feature vectors of a user, revealing vulnerabilities in face verification systems.

Findings

Generated images can match feature vectors but look different visually.

Face verification systems can be bypassed using generated images.

Vulnerabilities exist in real-world face verification applications.

Abstract

Face verification systems aim to validate the claimed identity using feature vectors and distance metrics. However, no attempt has been made to bypass such a system using generated images that are constrained by the same feature vectors. In this work, we train StarGAN v2 to generate diverse images based on a human user, that have similar feature vectors yet qualitatively look different. We then demonstrate a proof of concept on a custom face verification system and verify our claims by demonstrating the same proof of concept in a black box setting on dating applications that utilize similar face verification systems.