CVMay 29

SteerFace: Debiasing Synthetic Face Generation via Adaptive Residue Perturbation

arXiv:2605.3089494.11 citationsh-index: 13
AI Analysis

This paper tackles the problem of improving the utility of synthetic face data for face recognition systems, which is important for researchers and developers in computer vision and AI ethics.

This paper addresses the problem of a "synthetic-real gap" in face recognition performance when using diffusion-generated synthetic faces. They found that synthetic data has an unrealistic prevalence of visual attributes, which they call "visual tendency." They propose SteerFace, a training framework that perturbs identity embeddings to discourage the generator from relying on non-identity visual cues. This method effectively mitigates visual tendency, outperforms prior methods in downstream face recognition, and generalizes across different training datasets and generation pipelines.

The shortage of legally compliant data for face recognition training has sparked growing interest in using synthetic data as an alternative. While recent diffusion-based methods enable the generation of photorealistic face images with strong identity adherence and data diversity, their downstream recognition performance still exhibits a significant synthetic-real gap. This paper identifies visual tendency as a previously underexplored limitation, whereby synthetic data exhibit an unrealistic prevalence of visual attributes and thus deviate from the real-data distribution. Visual tendency can be attributed to the generator's conditioning on identity embeddings, through which co-occurring residual visual cues are unintentionally absorbed into learned identity semantics. To discourage the generator from exploiting such visual cues, this paper proposes SteerFace, a simple and efficient training framework that perturbs identity embeddings by steering them toward random orthogonal directions on the embedding hypersphere. The perturbation serves as an identity-preserving regularizer that penalizes the generator's reliance on non-identity components, as supported by theoretical analysis. This paper further introduces an adaptive strategy that learns perturbation strengths with both sample-wise preference and favorable overall statistics. Extensive experiments show that SteerFace effectively mitigates visual tendency, outperforms prior methods in downstream face recognition, and generalizes well across different training datasets and generation pipelines.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes