Multi-Modal Spoofing Attacks on 3D Face Liveness Detection via a Single 2D Photo

Face authentication technology has been widely used in physical access control to critical infrastructures. The security of a face authentication system has been threatened by photo replay attacks and thus the 3D liveness detection techniques have been deployed to safeguard such systems. In this paper, we conduct a comprehensive analysis of the security aspects pertaining to 3D liveness detection systems that employ structured light depth camera, and propose a novel attack surface targeting 3D face authentication systems involving multiple modalities such as Depth, RGB and IR. We propose the <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DepthFake</monospace> attack, a multi-modal spoofing attack against real-world 3D face authentication using only a single 2D photo. To achieve it, <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DepthFake</monospace> first reconstruct the depth information of the victim's face from his 2D photo. Then, <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DepthFake</monospace> actively projects a carefully-crafted scatter patterns embedded with the face depth information, in order to empower the 2D photo with 3D authentication properties. We address a range of practical challenges, including mitigating depth estimation errors, achieving depth images forgery techniques based on structured light, ensuring accurate alignment between various modalities of face images, and effectively implementing <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DepthFake</monospace> in real world. We validated <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DepthFake</monospace> on 5 commercial face authentication systems (i.e., Tencent Cloud, Baidu Cloud, 3DiVi, Ali Cloud and ArcSoft) and two commercial access control devices. The results over 50 users demonstrate that <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DepthFake</monospace> achieves an overall Depth attack success rate of 79.4%, RGB-D attack success rate of 59.4%, IR-D attack success rate of 79.4%, and RGB-IR attack success rate of 83.8% in the real world.