gsethia08/part1-clip_unclip_papers

这是一个用于机器人学的数据集，使用LeRobot框架创建。数据集包含403个训练集episodes，总计477,218帧视频数据，帧率为30fps，涵盖6个不同任务。数据特征包括：来自ego视角相机的RGB视频观测（分辨率1080x1920）、48维手部状态观测（包含左右手腕的6维位姿和每只手的5个手指的3维关节位置）、相机外参（16维）和内参（9维）、48维动作向量、语言指令文本以及时间戳、帧索引、episode索引等元数据。数据集主要用于机器人模仿学习和操作任务研究。This is a robotics dataset created using the LeRobot framework. It contains 403 training episodes with a total of 477,218 frames at 30fps, covering 6 different tasks. The dataset features include: RGB video observations from an ego-centric camera (resolution 1080x1920), 48-dimensional hand state observations (including 6D pose for both wrists and 3D joint positions for 5 fingers per hand), camera extrinsics (16D) and intrinsics (9D), 48-dimensional action vectors, language instruction text, and metadata such as timestamps, frame indices, and episode indices. The dataset is primarily intended for research in robot imitation learning and manipulation tasks.