Abstract/Summary

Face masks and coverings are often encountered by facial examiners (‘examiners’) in forensic case work. Examiners are skilled at unconcealed face identifications, but their accuracy for masked face identifications is unknown, yet can be used as evidence in court. Here we test performance of an international sample of 61 examiners, 39 professional teams, and 6 face identification algorithms for 20 image pairs. Pairs consisted of one unconcealed face image and one mask wearing face image. Examiners and professional teams outperformed controls, but professional teams made the least errors of all groups. The algorithms achieved high accuracy on the task. The findings back the notion that examiners use feature-based comparison strategies, and these are successful for matching images where one face wears a mask. Our results support the use of examiners for the identification of masked faces and suggest a role for teams and human-machine working in applied practice.