YTF

该数据集包含了为研究视频中无约束人脸验证问题而捕捉的人脸视频，共包含3,425个视频，涉及1,595位不同人士。此外，该数据集已经以帧的形式提供，并进行了面部检测和对齐的预处理。规模上，该数据集共有3,425个视频，任务是对视频中的人脸进行验证。

This dataset comprises facial videos captured for research on unconstrained face verification in video content, containing a total of 3,425 videos involving 1,595 distinct individuals. Furthermore, the dataset has been provided in the form of individual frames and has undergone preprocessing steps including face detection and alignment. In terms of scale, this dataset includes 3,425 videos, and the core task is to conduct face verification on the faces within these videos.