作者: Meiling Fang , Naser Damer , Florian Kirchbuchner , Arjan Kuijper
DOI: 10.23919/EUSIPCO47968.2020.9287321
关键词:
摘要: With the widespread use of biometric systems, demographic bias problem raises more attention. Although many studies addressed issues in verification, there are no works that analyze presentation attack detection (PAD) decisions. Hence, we investigate and iris PAD algorithms this paper. To enable a clear discussion, adapt notions differential performance outcome to problem. We study using three baselines (hand-crafted, transfer-learning, training from scratch) NDCLD-2013 [18] database. The experimental results point out female users will be significantly less protected by PAD, comparison males.