Enhancing Facial Data Diversity with Style-based Face Aging

TL;DR

This paper introduces a style-based generative model for augmenting face datasets with diverse age representations, improving fairness and performance in age-related face recognition tasks.

Contribution

It presents a novel architecture that captures detailed aging patterns for data augmentation, outperforming existing methods in age transfer and dataset diversity.

Findings

Outperforms state-of-the-art age transfer algorithms

Increases diversity in augmented datasets

Effective in both single- and cross-database evaluations

Abstract

A significant limiting factor in training fair classifiers relates to the presence of dataset bias. In particular, face datasets are typically biased in terms of attributes such as gender, age, and race. If not mitigated, bias leads to algorithms that exhibit unfair behaviour towards such groups. In this work, we address the problem of increasing the diversity of face datasets with respect to age. Concretely, we propose a novel, generative style-based architecture for data augmentation that captures fine-grained aging patterns by conditioning on multi-resolution age-discriminative representations. By evaluating on several age-annotated datasets in both single- and cross-database experiments, we show that the proposed method outperforms state-of-the-art algorithms for age transfer, especially in the case of age groups that lie in the tails of the label distribution. We further show significantly increased diversity in the augmented datasets, outperforming all compared methods according to established metrics.