CK+

该数据集包含了593个图像序列（视频），来自123位主体，每个序列展示了一种单一的面部表情。在实验中，我们使用了其中327个视频的子集，这些视频的标签已得到验证，对应于七种情绪。每个视频帧由136维特征表示构成，这些特征基于68个特征点位置的变化。规模上，共有327个视频序列，任务是对面部表情进行分类。

This dataset contains 593 image sequences (videos) sourced from 123 subjects, with each sequence presenting a single facial expression. For the conducted experiments, we utilized a subset of 327 videos from this dataset; the labels of these videos have been validated and correspond to seven emotional categories. Each video frame is represented by a 136-dimensional feature vector derived from the positional changes of 68 facial landmarks. In terms of scale, this experimental subset includes 327 video sequences in total, and the primary task is facial expression classification.