Literature review.

Existing algorithms for addressing disguise variations. AR database [33] contains 3200+ images pertaining to 126 subjects with two kinds of disguises (sunglasses and scarves). The National Geographic (NG) dataset contains 46 images of 1 individual, with various accessories such as hat, glasses, sunglasses, and facial hair. *Private dataset of 150 images pertaining to 15 individuals which contains similar real and synthetic disguise variations as in NG dataset. Synthetic disguise dataset of 4000 images pertaining to 100 individuals. Private datasets are collected by researches in real world scenarios from ATM (automatic teller machine) kiosk.