Differential Morphed Face Detection Using Deep Siamese Networks

Although biometric facial recognition systems are fast becoming part of security applications, these still vulnerable to morphing attacks, in which a reference image can be verified as two or more separate identities. In border control scenarios, successful attack allows people use the same passport cross borders. this paper, we propose novel differential morph detection framework using deep Siamese network. To best our knowledge, is first research work that makes network architecture for detection. We compare model with other classical and learning models distinct datasets, VISAPP17 MorGAN. explore embedding space generated by contrastive loss three decision making frameworks Euclidean distance, feature difference support vector machine classifier, concatenation classifier.