NegFaceDiff: The Power of Negative Context in Identity-Conditioned Diffusion for Synthetic Face Generation

TL;DR

NegFaceDiff introduces a negative conditioning sampling method in identity-conditioned diffusion models, significantly improving the separation and quality of synthetic face data for face recognition tasks.

Contribution

The paper presents NegFaceDiff, a novel negative conditioning sampling technique that enhances identity separation in synthetic face generation.

Findings

FDR increased from 2.427 to 5.687 with NegFaceDiff.

FR models trained on NegFaceDiff data outperform those trained on non-negative conditioned data.

NegFaceDiff improves identity consistency and separability in generated face images.

Abstract

The use of synthetic data as an alternative to authentic datasets in face recognition (FR) development has gained significant attention, addressing privacy, ethical, and practical concerns associated with collecting and using authentic data. Recent state-of-the-art approaches have proposed identity-conditioned diffusion models to generate identity-consistent face images, facilitating their use in training FR models. However, these methods often lack explicit sampling mechanisms to enforce inter-class separability, leading to identity overlap in the generated data and, consequently, suboptimal FR performance. In this work, we introduce NegFaceDiff, a novel sampling method that incorporates negative conditions into the identity-conditioned diffusion process. NegFaceDiff enhances identity separation by leveraging negative conditions that explicitly guide the model away from unwanted features while preserving intra-class consistency. Extensive experiments demonstrate that NegFaceDiff significantly improves the identity consistency and separability of data generated by identity-conditioned diffusion models. Specifically, identity separability, measured by the Fisher Discriminant Ratio (FDR), increases from 2.427 to 5.687. These improvements are reflected in FR systems trained on the NegFaceDiff dataset, which outperform models trained on data generated without negative conditions across multiple benchmarks.