An enhanced spatial-temporal graph convolution network with high order features for skeleton-based action recognition
Fig 5
(a) Illustrates the structure of the spatio-temporal block within the main branch, whereas (b) Provides a detailed description of the adaptive graph convolutional layer.