Synthetic Data Generation

该数据集通过一个综合的数据生成管道，生成了多样化且完美校准的图像数据，这些数据拥有像素级完美的标注，适用于训练高斯头像模型。该数据集包含了面部轮廓、纹理、服装、发型、发色和眼色的多样化变化，所有场景均采用均匀的白色光照进行照明。规模上，该数据集包含了1000个身份，每人50张图像。其任务是用以训练高斯头像模型。

This dataset generates diverse and perfectly calibrated image data through a comprehensive data generation pipeline, which has pixel-perfect annotations and is suitable for training Gaussian avatar models. The dataset includes diverse variations in facial contours, textures, clothing, hairstyles, hair colors and eye colors, and all scenes are illuminated by uniform white lighting. In terms of scale, the dataset contains 1000 identities, with 50 images per identity. The task of this dataset is to train Gaussian avatar models.