Animation Dataset

该数据集由超过一百万个原始动画视频构建而成，经过筛选保留了高质量的文本-视频对以供训练使用。数据集还包括根据文本覆盖区域、光流分数、审美分数和时长等标准过滤后的视频片段。规模上，该数据集包含了超过一千万个视频片段，旨在应对动画视频生成的任务。

This dataset is constructed from over one million raw animated videos, and high-quality text-video pairs have been retained through screening for training purposes. It also includes video clips filtered based on criteria such as text coverage area, optical flow score, aesthetic score and duration. With a total of more than 10 million video clips, this dataset is designed to address the task of animated video generation.