Analysis of the synthetic periocular iris images for robust Presentation Attacks Detection algorithms

The LivDet-2020 competition focuses on Presentation Attacks Detection (PAD) algorithms, has still open problems, mainly unknown attack scenarios. It is crucial to enhance PAD methods. This can be achieved by augmenting the number of Attack Instruments (PAI) and Bona fide (genuine) images used train such algorithms. Unfortunately, capture creation PAI even are sometimes complex achieve. generation synthetic with Generative Adversarial Networks (GAN) algorithms may help shown significant improvements in recent years. paper presents a benchmark GAN methods achieve novel from small set periocular near-infrared images. best was obtained using StyleGAN2, it tested algorithm LivDet-2020. able fool an algorithm. As result, all were classified as fide. A MobileNetV2 trained new class more robust PAD. resulting classify 96.7% attacks. BPCER10 0.24%. Such results demonstrated need for constantly updated