Research Question
How have the false positive error rates across age, sex, and race-based demographic groups in the Lumen facial recognition rank-one program evolved over the years?
Overview
Summary
Colorado law enforcement has increasingly incorporated facial recognition technologies to maximize efficiency. False positive match rates (FPMRs) are the rate at which facial recognition programs falsely associate one person with another. For my project, I will be analyzing the effectiveness of the Lumen rankone algorithm in regard to demographics - age, sex, and gender - through FPMRs.
Process
I will be using a variety of different equations derived from previously published works to develop a set of FPMRs per each of the six rankone versions by varying age, sex, and gender. After that, I will analyze this data to find trends in how the demographic accuracy improved from version to version.
