CemiFace: Center-based Semi-hard Synthetic Face Generation for Face Recognition

TL;DR

CemiFace introduces a diffusion-based method to generate synthetic face images with controlled similarity to identity centers, enhancing face recognition training while addressing privacy concerns.

Contribution

The paper proposes a novel center-based semi-hard synthetic face generation approach that improves discriminative quality of training data for face recognition.

Findings

Generated datasets with moderate similarity improve recognition performance.

CemiFace outperforms existing synthetic face generation methods.

Training with CemiFace data achieves competitive accuracy.

Abstract

Privacy issue is a main concern in developing face recognition techniques. Although synthetic face images can partially mitigate potential legal risks while maintaining effective face recognition (FR) performance, FR models trained by face images synthesized by existing generative approaches frequently suffer from performance degradation problems due to the insufficient discriminative quality of these synthesized samples. In this paper, we systematically investigate what contributes to solid face recognition model training, and reveal that face images with certain degree of similarities to their identity centers show great effectiveness in the performance of trained FR models. Inspired by this, we propose a novel diffusion-based approach (namely Center-based Semi-hard Synthetic Face Generation (CemiFace)) which produces facial samples with various levels of similarity to the subject center, thus allowing to generate face datasets containing effective discriminative samples for training face recognition. Experimental results show that with a modest degree of similarity, training on the generated dataset can produce competitive performance compared to previous generation methods.