2020/10/25 by Mehak Gupta, Vishal Singh, Gupta, Mehak +7
Computer Science · #Biometric Identification and Security #Computer Vision and Pattern Recognition (cs.CV) #Cryptography and Security (cs.CR) #FOS: Computer and information sciences #Face recognition and analysis
paper · pdf · doi:10.48550/arxiv.2010.13244
openalex publication_date 2020/10/25 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Presentation attacks are posing major challenges to most of the biometric\nmodalities. Iris recognition, which is considered as one of the most accurate\nbiometric modality for person identification, has also been shown to be\nvulnerable to advanced presentation attacks such as 3D contact lenses and\ntextured lens. While in the literature, several presentation attack detection\n(PAD) algorithms are presented; a significant limitation is the\ngeneralizability against an unseen database, unseen sensor, and different\nimaging environment. To address this challenge, we propose a generalized deep\nlearning-based PAD network, MVANet, which utilizes multiple representation\nlayers. It is inspired by the simplicity and success of hybrid algorithm or\nfusion of multiple detection networks. The computational complexity is an\nessential factor in training deep neural networks; therefore, to reduce the\ncomputational complexity while learning multiple feature representation layers,\na fixed base model has been used. The performance of the proposed network is\ndemonstrated on multiple databases such as IIITD-WVU MUIPA and IIITD-CLI\ndatabases under cross-database training-testing settings, to assess the\ngeneralizability of the proposed algorithm.\n