未明确命名

本研究旨在创建一个平衡的人脸数据集，以减少人脸识别模型中的偏见。数据集通过使用StyleGAN模型生成，能够根据不同的群体特征（如性别、种族）生成合成图像。该数据集的创建过程涉及使用StyleGAN的潜空间进行解耦，以便能够有效地生成任何群体组合的图像。数据集主要应用于人脸识别领域，旨在解决现有数据集中群体不平衡导致的识别偏差问题。

This study aims to create a balanced face dataset to mitigate bias in facial recognition models. This dataset is generated using the StyleGAN model, which can synthesize images based on different demographic characteristics such as gender and ethnicity. The creation process of this dataset involves disentangling the latent space of StyleGAN, enabling efficient generation of images with any combination of demographic groups. This dataset is primarily applied in the field of facial recognition, aiming to address recognition bias issues caused by group imbalance in existing datasets.