Demographic Variability in Face Image Quality Measures
Wassim Kabbani, Kiran Raja, Raghavendra Ramachandra, Christoph Busch
TL;DR
This work addresses potential demographic differential in face image quality assessment (FIQA) measures defined by ISO/IEC 29794-5 across age, gender, and skin tone. It evaluates all FIQA measures on four datasets using the Monk Skin Tone Scale (MST) for skin tone labeling and leverages the OFIQ implementation of the standard, with auxiliary FAC-based labeling to fill gaps. The findings show that most measures exhibit no clear demographic differential, but two skin-tone-related measures—dynamic range and luminance mean—display notable variation with skin tone, suggesting targeted bias. The results support the cautious deployment of FIQA in online identity management, while identifying specific measures requiring mitigation to ensure fairness across diverse populations.
Abstract
Face image quality assessment (FIQA) algorithms are being integrated into online identity management applications. These applications allow users to upload a face image as part of their document issuance process, where the image is then run through a quality assessment process to make sure it meets the quality and compliance requirements. Concerns about demographic bias have been raised about biometric systems, given the societal implications this may cause. It is therefore important that demographic variability in FIQA algorithms is assessed such that mitigation measures can be created. In this work, we study the demographic variability of all face image quality measures included in the ISO/IEC 29794-5 international standard across three demographic variables: age, gender, and skin tone. The results are rather promising and show no clear bias toward any specific demographic group for most measures. Only two quality measures are found to have considerable variations in their outcomes for different groups on the skin tone variable.