Skeleton Action Recognition Method Based on the Combination of Dynamic and Static Diagram
SHU Yang
LI Wanggen
GAO Kun
WANG Zhige
GE Yingkui
Abstract：With the further development of graph convolution neural network,the method of human skeleton motion recogni-tion based on graph convolution network is becoming popular.However,most of the current graph convolution methods rely too much on the fixed joint connection relationship on the skeleton,ignoring the potential relationship between non adjacent joints in time and space,and taking up a lot of computing costs.To solve the above problems,this paper proposes a skeleton motion recogni-tion method based on the combination of dynamic and static diagrams(DS-GCN).This paper combines the dynamic channel map transformed by semantic information with the static skeleton map of human body,and achieves good performance under low parame-ter load.
Keywords：graph convolution neural networkaction recognitiondynamic channel diagramstatic skeleton diagramlow parameter
Publication Date：2025-06-20
Online Publishing Date：2025-09-23(First online date of this platform, not the publication date of the document)
Pages：6( 1585-1590 )