ROI-su7/act_pick

该数据集使用LeRobot创建，专注于机器人任务，可能涉及与物体的操作或交互。数据集结构包括动作、观察状态、手眼图像、固定图像等多种特征，以及相关的数据类型、形状等信息。数据集仅包含训练数据，没有测试或验证数据集的明确提及。提供了总剧集数、总帧数和任务数，以及数据和视频文件的大小。数据集组织有序，有专门的数据和视频文件路径，使用Parquet文件格式存储数据，视频文件为MP4格式，具有特定的编解码器和像素格式。

This dataset, created using LeRobot, focuses on robotic tasks, possibly involving manipulation or interaction with objects. The dataset structure includes features like action, observation.state, handeye images, fixed images, etc., along with their data types, shapes, and other related information. The dataset contains only training data, with no explicit mention of test or validation splits. It provides the total number of episodes, frames, and tasks, as well as the size of the data and video files. The dataset is well-organized, with specific paths for data and video files, using the Parquet file format for data storage and MP4 format for video files with specific codec and pixel format details.