标题： Vec2Face+ 用于人脸数据集生成 显示英文标题

标题： Vec2Face+ for Face Dataset Generation

摘要： 当将身份合成作为人脸识别的训练数据时，通常认为大的类间可分性和类内属性变化对于合成高质量的数据集是必不可少的。 % 这种信念通常是正确的，这也是我们的目标。 然而，当增加类内变化时，现有方法忽略了保持类内身份一致性的必要性。 % 为了解决这个问题并生成高质量的人脸训练数据，我们提出了Vec2Face+，这是一种生成模型，可以直接从图像特征生成图像，并允许对人脸身份和属性进行连续且容易的控制。 使用Vec2Face+，我们通过三种策略获得具有适当类间可分性、类内变化和身份一致性的数据集：1）我们采样与其它向量足够不同的向量以生成良好分离的身份；2）我们提出了一种AttrOP算法来增加一般属性的变化；3）我们提出了基于LoRA的姿态控制，用于生成具有侧面头部姿势的图像，这比AttrOP更高效且保留身份。 % 我们系统生成了VFace10K，一个包含10K身份的合成人脸数据集，使得FR模型在七个真实世界测试集中达到最先进的准确率。 将规模扩展到4M和12M图像，相应的VFace100K和VFace300K数据集在五个真实世界测试集中比真实世界训练数据集CASIA-WebFace具有更高的准确率。 这是第一次合成数据集在平均准确率上超过了CASIA-WebFace。 此外，我们发现只有1个合成数据集在双胞胎验证中优于随机猜测（\emph{即，50%}），并且使用合成身份训练的模型比使用真实身份训练的模型更具偏差。 这些都是未来研究的重要方面。 显示英文摘要

摘要： When synthesizing identities as face recognition training data, it is generally believed that large inter-class separability and intra-class attribute variation are essential for synthesizing a quality dataset. % This belief is generally correct, and this is what we aim for. However, when increasing intra-class variation, existing methods overlook the necessity of maintaining intra-class identity consistency. % To address this and generate high-quality face training data, we propose Vec2Face+, a generative model that creates images directly from image features and allows for continuous and easy control of face identities and attributes. Using Vec2Face+, we obtain datasets with proper inter-class separability and intra-class variation and identity consistency using three strategies: 1) we sample vectors sufficiently different from others to generate well-separated identities; 2) we propose an AttrOP algorithm for increasing general attribute variations; 3) we propose LoRA-based pose control for generating images with profile head poses, which is more efficient and identity-preserving than AttrOP. % Our system generates VFace10K, a synthetic face dataset with 10K identities, which allows an FR model to achieve state-of-the-art accuracy on seven real-world test sets. Scaling the size to 4M and 12M images, the corresponding VFace100K and VFace300K datasets yield higher accuracy than the real-world training dataset, CASIA-WebFace, on five real-world test sets. This is the first time a synthetic dataset beats the CASIA-WebFace in average accuracy. In addition, we find that only 1 out of 11 synthetic datasets outperforms random guessing (\emph{i.e., 50\%}) in twin verification and that models trained with synthetic identities are more biased than those trained with real identities. Both are important aspects for future investigation.