Fig. 3From: STTG-net: a Spatio-temporal network for human motion prediction based on transformer and graph convolution networkThe architecture of S-GCN moduleBack to article page