ACTIVE（Action from Robotic View）是一个专为移动服务机器人感知视角设计的大规模数据集，包含30个复合动作类别、80名参与者和46,868个注释视频实例，涵盖RGB和点云两种模态。参与者在不同环境中执行各种人类动作，距离从3米到50米不等，同时摄像机平台也是移动的，模拟了机器人感知的真实场景。该数据集旨在推动自然人机交互（N-HRI）中的动作和属性识别研究。
Hand contact data, reflecting the intricate behaviours of human hands during object operation, exhibits significant potential for analysing hand operation patterns to guide the design of hand-relate
Experimental data associated with the project and paper WhichFingers. https://dx.doi.org/10.1145/3126594.3126619 http://ns.inria.fr/mjolnir/whichfingers/
