Who is behind the Mask: Periocular Biometrics when Face Recognition Fails

The ongoing COVID-19 epidemic has emphasized the importance of hygienic and contactless identification au-thentication. On the other hand, the pandemic resulted in the widespread usage of face masks, which are still necessary to keep the infection under control. The consequence of wearing a mask on facial recognition is a sensitive yet understudied issue right now. Furthermore, the performance and reliability of contactless identity verification using facial recognition can indeed be affected by different types, sizes, and colors of masks. Therefore, this study evaluates the performance of face recognition system with various kinds of occlusions. Periocular region based authentication, especially when only a partial face is available, is proposed. Experimental results prove the reliability, robustness, and superiority of periocular recognition over the face under studied circumstances.