Vulnerability Analysis of Face Morphing Attacks from Landmarks and Generative Adversarial Networks

Morphing attacks is a threat to biometric systems where the biometric\nreference in an identity document can be altered. This form of attack presents\nan important issue in applications relying on identity documents such as border\nsecurity or access control. Research in face morphing attack detection is\ndeveloping rapidly, however very few datasets with several forms of attacks are\npublicly available. This paper bridges this gap by providing a new dataset with\nfour different types of morphing attacks, based on OpenCV, FaceMorpher,\nWebMorph and a generative adversarial network (StyleGAN), generated with\noriginal face images from three public face datasets. We also conduct extensive\nexperiments to assess the vulnerability of the state-of-the-art face\nrecognition systems, notably FaceNet, VGG-Face, and ArcFace. The experiments\ndemonstrate that VGG-Face, while being less accurate face recognition system\ncompared to FaceNet, is also less vulnerable to morphing attacks. Also, we\nobserved that na\\"ive morphs generated with a StyleGAN do not pose a\nsignificant threat.\n