Robot-Human Paired Pose Dataset

该数据集包含了200万个机器人与人类姿态配对的数据，这些数据是通过一种机器人至人类的姿态配对数据收集方法收集而来的，旨在用于训练一个两阶段的神经网络，以实现动作重定向。该数据集涵盖了7位不同主体执行的11种不同动作，其设计目的是支持使用全监督学习方法训练模型。

This dataset contains 2 million pairs of robot and human pose data, collected using a robot-to-human pose matching data collection methodology. It is developed for training a two-stage neural network to enable action redirection. The dataset encompasses 11 distinct actions performed by 7 different subjects, and is specifically designed to support model training via fully supervised learning approaches.