DAGroup-PKU/RoVid-X
收藏Hugging Face2026-01-25 更新2026-02-07 收录
下载链接：
https://hf-mirror.com/datasets/DAGroup-PKU/RoVid-X
下载链接
链接失效反馈官方服务：
资源简介：
RoVid-X是一个大规模机器人视频生成数据集，包含4百万个机器人视频片段（超过1万小时），涵盖1300多种细粒度的机器人技能。数据集提供多模态物理注释（包括RGB、深度和光流），支持多机器人和多任务多样性，涉及多种机器人类型、场景和动作技能。数据集结构以JSON格式提供，每个视频片段都有详细的注释，如动词、任务描述、简短描述和详细描述。
RoVid-X is a large-scale robotic video generation dataset containing 4M robotic video clips (10K+ hours) with 1300+ fine-grained robotic skills. It provides multi-modal physical annotations, including RGB, depth, and optical flow, and supports multi-robot and multi-task diversity across various robot types, scenarios, and action skills. The dataset is structured in JSON format, with detailed annotations for each video clip, such as verb, task caption, short caption, and detailed caption.
提供机构：
DAGroup-PKU
