bot-pi/tokyo_u_lsmo

该数据集是一个机器人数据集，包含50个视频，每个视频包含多个片段，每个片段包含1000帧。数据集总共包含11925帧，分为2个任务。数据集提供了视频帧、语言指令、机器人状态、动作、时间戳、奖励和完成状态等信息。视频帧的分辨率为120x120，帧率为5fps，格式为av1编码的MP4文件，不包含音频。数据集的结构化数据存储为Parquet格式。

This dataset is a robotics dataset containing 50 videos, each with multiple chunks, and each chunk containing 1000 frames. The dataset includes a total of 11,925 frames across 2 tasks. It provides information such as video frames, language instructions, robot states, actions, timestamps, rewards, and done status. The video frames are of resolution 120x120, with a frame rate of 5fps, encoded in MP4 format using av1 codec, and do not include audio. The structured data in the dataset is stored in Parquet format.