lerobot/nyu_door_opening_surprising_effectiveness
收藏Hugging Face2025-02-20 更新2025-04-08 收录
下载链接：
https://hf-mirror.com/datasets/lerobot/nyu_door_opening_surprising_effectiveness
下载链接
链接失效反馈官方服务：
资源简介：
VINN数据集是一个用于视觉模仿表示学习的机器人数据集，包含484个视频片段，每个片段包含1000帧，总共有20405帧。数据集使用LeRobot创建，所有视频均为3fps的AV1编码，格式为yuv420p。数据集包含多种特征，包括720x960的彩色图像、字符串类型的语言指令、机器人状态（包括8个电机的位置）、动作（控制7个电机）、时间戳、帧索引、剧集索引、下一帧的奖励和完成标志等。
The VINN dataset is a robotics dataset for visual imitation representation learning, containing 484 video chunks with 1000 frames each, totaling 20,405 frames. The dataset was created using LeRobot, with all videos encoded in AV1 format at 3fps and in yuv420p resolution. The dataset includes various features such as 720x960 color images, string-type language instructions, robot states (positions of 8 motors), actions (controlling 7 motors), timestamps, frame indices, episode indices, next frame rewards, and done flags.
提供机构：
lerobot
