Probing Fairness of Mobile Ocular Biometrics Methods Across Gender on VISOB 2.0 Dataset

Recent research has questioned the fairness of face-based recognition and attribute classification methods (such as gender race) for dark-skinned people women. Ocular biometrics in visible spectrum is an alternate solution over face biometrics, thanks to its accuracy, security, robustness against facial expression, ease use mobile devices. With recent COVID-19 crisis, ocular a further advantage presence mask. However, not been studied till now. This first study aims explore ocular-based authentication across males females. To this aim, VISOB 2.0 dataset, along with annotations, used analysis based on ResNet-50, MobileNet-V2 lightCNN-29 models. Experimental results suggest equivalent performance females user-authentication terms genuine match rate (GMR) at lower false rates (FMRs) overall Area Under Curve (AUC). For instance, AUC 0.96 0.95 was obtained average. significantly outperformed deep learning models ocular-region.