Articles
| Open Access | CRAFTING DUAL-IDENTITY FACE IMPERSONATIONS USING GENERATIVE ADVERSARIAL NETWORKS: AN ADVERSARIAL ATTACK METHODOLOGY
Abstract
With the rapid advancement of face recognition systems, security threats posed by adversarial attacks have become increasingly sophisticated. This study presents a novel adversarial methodology for crafting dual-identity face impersonations using Generative Adversarial Networks (GANs). The proposed framework generates synthetic facial images that simultaneously resemble two distinct target identities, thereby enabling high-confidence impersonation across multiple recognition systems. Leveraging a multi-objective loss function, the generator is trained to optimize both identity similarity scores and realism metrics while evading detection from spoofing and liveness classifiers. Extensive evaluations on benchmark datasets such as LFW and CASIA-WebFace demonstrate the effectiveness of the method in deceiving state-of-the-art face verification models with minimal perceptual distortion. The research highlights the vulnerabilities of current biometric systems and underscores the urgent need for robust defense mechanisms against such dual-target adversarial threats.
Keywords
Face impersonation, dual-identity attack, generative adversarial networks (GANs), biometric spoofing
References
Article Statistics
Downloads
Copyright License
Authors retain the copyright of their manuscripts, and all Open Access articles are disseminated under the terms of the Creative Commons Attribution License 4.0 (CC-BY), which licenses unrestricted use, distribution, and reproduction in any medium, provided that the original work is appropriately cited. The use of general descriptive names, trade names, trademarks, and so forth in this publication, even if not specifically identified, does not imply that these names are not protected by the relevant laws and regulations.