This article describes a statistical analysis that quantifies the effects of covariates on three of the better performing algorithms in the Face Recognition Vendor Test 2006, with the goals of identifying which covariates impact algorithm performance and quantifying those effects.
A study is presented showing how three state-of-the-art algorithms from the Face Recognition Vendor Test 2006 (FRVT 2006) are affected by factors related to face images and people. The recognition scenario compares highly controlled images to images taken of people as they stand before a camera in settings such as hallways and outdoors in front of buildings. A Generalized Linear Mixed Model (GLMM) is used to estimate the probability an algorithm successfully verifies a person conditioned upon the factors included in the study. The factors associated with people are: Gender, Race, Age, and whether or not they wear Glasses. The factors associated with images are: the size of the face, edge density and region density. The setting, indoors versus outdoors, is also a factor. Edge density can change the estimated probability of verification dramatically, for example from about 0.15 to 0.85. However, this effect is not consistent across algorithm or setting. This finding shows that simple measurable factors are capable of characterizing face quality; however, these factors typically interact with both algorithm and setting. (Published Abstract Provided)
Downloads
Similar Publications
- Metal pollution investigation of Goldman Park, Middletown Ohio: Evidence for steel and coal pollution in a high child use setting
- Application of X-ray photoelectron spectroscopy to examine surface chemistry of cancellous bone and medullary contents to refine bone sample selection for nuclear DNA analysis
- Reflective spectroscopy investigations of clothing items to support law enforcement, search and rescue, and war crime investigations