VGGFace2

该数据集提供了类似于MultiLab的姿势模板，并包含了测试集中的一部分图像。此外，该数据集还包含了针对各种姿势标注的图像。其规模包括10,750张图像（测试集）和63,016张图像（训练集），所涉及的任务是面部表情识别。

This dataset provides pose templates similar to those employed in MultiLab, and includes a subset of images from the test set. Furthermore, it contains images annotated with various pose labels. The dataset consists of 10,750 images in the test set and 63,016 images in the training set, with the targeted task being facial expression recognition.