The Emperor's New Masks: On Demographic Differences and Disguises

Katherine L. Gibson, Jonathan M. Smith; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2015, pp. 57-64

Remaining unrecognized in an era of ubiquitous camera surveillance remains desirable to some, but advances in face recognition technology make it increasingly difficult to do so. A large database of high-quality imagery was used to explore the effectiveness of disguise as an approach to avoiding recognition. A commercial system that was highly rated in NIST's Face Recognition Vendor Test was used to evaluate a variety of disguises worn by each member of a study population that was diverse in age, gender, and race. Analysis of the recognition results for subsets extracted from the population shows that disguise can be remarkably effective. However, the efficacy of the disguises against face recognition varies so significantly with demographics that, for some, the disguises are not worth wearing.