Beegbrain/libero_slots_nd

该数据集是一个用于机器人学习的数据集，由LeRobot创建。它包含使用panda机器人采集的1693个episodes，总计273,465帧，覆盖40个不同任务。数据以parquet文件格式存储，视频以mp4格式存储，帧率为10fps。数据集特征包括两个分辨率为256x256的RGB图像观测（image和image2）、一个8维的状态观测向量、一个7维的动作向量，以及时间戳、帧索引、episode索引、任务索引等元数据。此外，还包含图像槽位（slots）和掩码（masks）特征，用于高级视觉处理。所有数据仅用于训练分割，适用于机器人控制、强化学习或计算机视觉任务的研究。This dataset is a robotics learning dataset created using LeRobot. It contains 1,693 episodes collected with a panda robot, totaling 273,465 frames and covering 40 different tasks. The data is stored in parquet format, with videos in mp4 format at a frame rate of 10fps. Dataset features include two RGB image observations (image and image2) at 256x256 resolution, an 8-dimensional state observation vector, a 7-dimensional action vector, and metadata such as timestamp, frame index, episode index, and task index. Additionally, it includes image slots and masks features for advanced visual processing. All data is allocated for the training split, making it suitable for research in robot control, reinforcement learning, or computer vision tasks.