Voxel51/motif-1k

MotIF-1K 是一个机器人运动数据集，包含1,022个演示，覆盖13个任务类别，用于基准测试和微调视觉-语言模型（VLMs）以进行基于运动的成功检测。每个演示包括运动视频、多个预渲染的轨迹可视化、任务指令和运动描述。该数据集是一个分组数据集，每个组代表一个轨迹，每个组切片代表该轨迹的不同视觉表示——与论文中使用的输入格式完全一致。数据集由人类运动和Stretch机器人运动两个配置组成，其中部分演示因不完整而缺少预渲染可视化。数据集还包含坐标系统、标签字段和数据质量问题等详细信息。MotIF-1K is a robotics motion dataset containing 1,022 demonstrations across 13 task categories, used to benchmark and fine-tune vision-language models (VLMs) for motion-based success detection. Each demonstration includes a video of the motion, multiple pre-rendered trajectory visualizations, task instructions, and motion descriptions. The FiftyOne dataset is a grouped dataset where each group represents one trajectory and each group slice represents a different visual representation of that trajectory — mirroring the exact input formats used in the paper. The dataset consists of human motion and stretch motion configurations, with some incomplete demonstrations lacking pre-rendered visualizations. It also includes details on coordinate systems, label fields, and known data quality issues.