Examining the examiners: an online eyebrow verification experiment inspired by FISWG

Abstract

In forensic face comparison, one of the features taken into account are the eyebrows. In this paper, we investigate human performance on an eyebrow verification task. This task is executed twice by participants: a "best-effort" approach and an approach using features based on forensic knowledge. The group of participants is divided into forensic/biometric experts and non-experts. The rationale behind this experiment is to determine whether there exist differences between (a) experts and non-experts and (b) the "best-effort" and forensic approach. It is shown empirically on a special constructed dataset that there do not exist major differences between experts and non-experts, however non-experts do perform worse when using the forensic approach instead of the "best-effort" approach.