Body part-based dual-task spatio-temporal graph convolutional network for action quality assessment
A Body Part-Based Dual-Task Spatio-Temporal Graph Convolutional Network (BPDT-STGCN) is proposed, which partitioned the human skeleton into three anatomical parts, the head, hand, and leg, which are individually fed into Spatio-Temporal Graph Convolutional Networks (ST-GCN) for feature extraction.