YUP++

该数据集包含了20个场景类别的视频纹理，每个类别有60个视频，这些视频分别由静止或移动的摄像机拍摄而成。在此研究中，我们采用了标准的数据集划分方式，其中1/9的数据集用于训练。数据集规模属于小型，所面临的任务是场景识别。

This dataset comprises video textures spanning 20 scene categories, with 60 videos per category. All videos are captured using either stationary or moving cameras. In this study, we adopt the standard dataset splitting protocol, where 1/9 of the total dataset is allocated for training. This is a small-scale dataset, and the targeted task is scene recognition.