Human Motion and Expression Dataset

该数据集是一个由多种来源收集的视频数据构建的综合性数据集，总时长达到500小时。它包含了丰富的场景，如舞蹈、体育、电影片段和演讲等。在画面构图上，该数据集做到了均衡，全身镜头和半身镜头各占大约一半的数据。这个数据集的规模达到了500小时的视频素材，其任务旨在进行人体图像动画与动作合成。This is a comprehensive video dataset built from multi-source collected video materials, with a total duration of 500 hours. It covers a wide range of scenarios such as dance, sports, movie clips, and speeches. In terms of shot composition, the dataset maintains a balanced split, with full-body shots and half-body shots each making up roughly half of the total data. With a scale of 500 hours of video footage, this dataset is designed for human image animation and motion synthesis tasks.