天津科技大学学报2023，Vol.38Issue(6)：1-11,11.DOI:10.13364/j.issn.1672-6510.20230056
基于图卷积神经网络的人体骨架动作识别研究进展
Review of Human Skeleton Action Recognition Based on Graph Convolution Neural Network
摘要
Abstract
Action recognition based on human skeleton is an important branch to achieve computer vision intelligence.In this article we first review the research on human skeleton action recognition based on graph convolutional neural networks and analyze the key techniques.Then,we outline the research progress of graph convolution approaches based on spectral convolution and space domain convolution,and detail the research progress of graph convolution models in the field of hu-man skeleton action recognition from two perspectives of adjacency matrix and input features.Furthermore,we analyze and compare the existing algorithms for human skeleton action recognition based on graph convolution neural networks.Finally,we look forward to the future development direction of graph convolution neural networks in the field of human skeleton action recognition.关键词
图理论/图神经网络/图卷积神经网络/基于骨架的动作识别/时空域融合Key words
graph theory/graph neural network/graph convolutional neural network/skeleton-based action recogni-tion/spatiotemporal fusion分类
信息技术与安全科学引用本文复制引用
杨巨成,张泉钰,王波,王嫄,陈亚瑞,赵婷婷..基于图卷积神经网络的人体骨架动作识别研究进展[J].天津科技大学学报,2023,38(6):1-11,11.基金项目
国家自然科学基金项目(61976156) （61976156）
