finetrainers/OpenVid-10k-split
收藏Hugging Face2025-04-10 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/finetrainers/OpenVid-10k-split
下载链接
链接失效反馈官方服务：
资源简介：
OpenVid-1M是一个用于文本到视频任务的数据集，包含视频、字幕、美感评分、运动评分、时间一致性评分、相机运动类型、帧数、帧率、视频时长和部分ID等信息。这个数据集的子集包含了大约10k个视频，是为了在测试时能够更快地迭代。视频的筛选基于美感评分和运动评分，只包含帧数在80到240之间的视频。
OpenVid-1M is a text-to-video dataset containing video, caption, aesthetic score, motion score, temporal consistency score, camera motion type, frame count, frame rate, video duration, and part ID. This subset of the dataset includes about 10k videos, designed for faster iteration during testing. The videos are filtered based on aesthetic and motion scores, including only those with between 80 and 240 frames.
提供机构：
finetrainers
