Synthetic Handover Motions

该数据集是通过一种手物合成方法生成的合成人类交递动作数据集，涵盖了右手和左手的动作。测试集包含了通过镜像右手动作生成的左手动作，并允许在未见过的物体上进行评估。具体规模为：训练集包含1175个物体和2230个右手动作；测试集包含1174个物体和4436个动作。该数据集的任务是针对人向机器人交递物品的动作研究。This is a synthetic human object handover action dataset generated using a manual object synthesis approach, covering actions performed by both the right and left hands. The test set consists of left-hand actions generated by mirroring right-hand movements, allowing evaluation on previously unseen objects. In terms of dataset scale, the training set includes 1175 objects and 2230 right-hand actions, while the test set contains 1174 objects and 4436 total actions. The core task of this dataset is to support research on human-to-robot object handover behaviors.