FaceWarehouse

该数据集包含了代表9个身份的180个网格样本，每个身份都有20种不同的表情。其他信息中提到，性能报告是基于每个顶点的误差来衡量的。规模上，数据集包含180个网格，任务则是评估在3D人脸重建中表情表现的效果。

This dataset includes 180 mesh samples representing 9 unique identities, with 20 distinct expressions per identity. Supplementary information states that performance reports are evaluated using per-vertex error metrics. In terms of scale, the dataset contains 180 meshes, and the task is to assess the effectiveness of facial expression reconstruction in 3D face reconstruction.