GHCLNet: A Generalized Hierarchically tuned Contact Lens detection Network

摘要

Iris serves as one of the best biometric modality owing to its complex,unique and stable structure. However, it can still be spoofed using fabricatedeyeballs and contact lens. Accurate identification of contact lens is must forreliable performance of any biometric authentication system based on thismodality. In this paper, we present a novel approach for detecting contact lensusing a Generalized Hierarchically tuned Contact Lens detection Network(GHCLNet) . We have proposed hierarchical architecture for three class oculusclassification namely: no lens, soft lens and cosmetic lens. Our networkarchitecture is inspired by ResNet-50 model. This network works on raw inputiris images without any pre-processing and segmentation requirement and this isone of its prodigious strength. We have performed extensive experimentation ontwo publicly available data-sets namely: 1)IIIT-D 2)ND and on IIT-K data-set(not publicly available) to ensure the generalizability of our network. Theproposed architecture results are quite promising and outperforms the availablestate-of-the-art lens detection algorithms.