AI Evasion and Impersonation Attacks on Facial Re-Identification with Activation Map Explanations

Researchers developed a novel framework for generating adversarial patches that can fool facial recognition systems through both evasion and impersonation attacks. The method reduces facial recognition accuracy from 90% to 0.4% in white-box settings and demonstrates strong cross-model generalization, highlighting critical vulnerabilities in surveillance systems.

• →New adversarial patch framework can generate attacks in a single forward pass without iterative optimization for each target.
• →Evasion attacks reduce facial recognition mean Average Precision from 90% to 0.4% in white-box settings and 72% to 0.4% in black-box settings.
• →Impersonation attacks achieve 27% success rate on CelebA-HQ dataset, competing with existing patch-based methods.
• →The framework shows strong cross-model generalization, indicating widespread vulnerability across different facial recognition systems.
• →Researchers used activation map clustering to identify features exploited by attacks and propose pathways for future countermeasures.