Forged presentation attack detection for ID cards on remote verification systems
Sebastian Gonzalez, Juan Tapia
Abstract
This paper introduces a new hybrid two-stage multiclass network for end-to-end Presentation Attack Detection in remote biometric verification systems using ID cards based on MobileNetV2. Using different capture sources, several presentation attacks species such as printing, display, composite (based on cropped and spliced areas), plastic (PVC), and synthetic ID card images. This proposal was developed using a database consisting of 190,000 real-case Chilean ID card images with the support of a third-party company. Also, a new framework called PyPAD, used to estimate multi-class metrics compliant with the ISO/IEC 30107-3 standard, was developed and will be made available for research purposes. Our method is trained on two convolutional neural networks separately, reaching BPCER100 scores on ID card attacks of 1.69% for network 1 and 2.36% for network 2. Using both models, The two-stage concatenated method can achieve a BPCER 100 score of 0.92%. • This work is one of the first end-to-end PAD evaluations on ID cards. • A serial, two-stage architecture trained from scratch is proposed. • New fake images are included based on synthetic and PVC ID cards. • This paper presents results using an ID Card database of 190K images. • A new toolkit for the assessment of multi-class PAD systems is proposed.