IDperturb：通过角度扰动增强合成人脸生成的多样性
IDperturb: Enhancing Variation in Synthetic Face Generation via Angular Perturbations
对做人脸识别训练数据的人：把身份嵌入限制在单位超球面的小角度锥内做扰动采样，无需重训扩散模型即可提升合成人脸的类内多样性，可直接作为身份条件扩散生成器的即插即用采样策略。
原文依据 IDPERTURB perturbs identity embeddings within a constrained angular region of the unit hyper-sphere
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论合成数据已成为训练人脸识别（FR）系统时替代真实人脸数据集的实用方案，尤其是在隐私与法律顾虑日益限制真实生物特征数据使用的情况下。身份条件扩散模型的最新进展使得生成照片级真实且身份一致的人脸图像成为可能。然而，这类模型中的许多都存在类内变化有限的问题，而类内变化是训练鲁棒且可泛化的 FR 模型所必需的性质。在本工作中，我们提出 IDPERTURB，一种简单而有效的几何驱动采样策略，用于增强合成人脸生成中的多样性。IDPERTURB 在单位超球面的受限角度区域内对身份嵌入进行扰动，在不修改底层生成模型的情况下产生一组多样化的嵌入。每个被扰动的嵌入都作为预训练扩散模型的条件向量，从而能够合成视觉上多样但身份连贯的人脸图像，适用于训练可泛化的 FR 系统。实证结果表明，与现有的合成数据生成方法相比，使用 IDPERTURB 生成的数据集训练 FR 在多个 FR 基准上取得了更优的性能。代码与生成的数据集已公开于 https://github.com/fdbtrs/IDperturb。
英文原文摘要 1345 字符
Synthetic data has emerged as a practical alternative to authentic face datasets for training face recognition (FR) systems, especially as privacy and legal concerns increasingly restrict the use of real biometric data. Recent advances in identity-conditional diffusion models have enabled the generation of photorealistic and identity-consistent face images. However, many of these models suffer from limited intra-class variation, an essential property for training robust and generalizable FR models. In this work, we propose IDPERTURB, a simple yet effective geometric-driven sampling strategy to enhance diversity in synthetic face generation. IDPERTURB perturbs identity embeddings within a constrained angular region of the unit hyper-sphere, producing a diverse set of embeddings without modifying the underlying generative model. Each perturbed embedding serves as a conditioning vector for a pre-trained diffusion model, enabling the synthesis of visually varied yet identity-coherent face images suitable for training generalizable FR systems. Empirical results demonstrate that training FR on datasets generated using IDPERTURB yields improved performance across multiple FR benchmarks, compared to existing synthetic data generation approaches. Code and generated datasets are publicly available https://github.com/fdbtrs/IDperturb.