Skip to main navigation Skip to search Skip to main content

Hierarchical Graph Convolutional Network for Skeleton-Based Action Recognition

  • Linjiang Huang
  • , Yan Huang
  • , Wanli Ouyang
  • , Liang Wang*
  • *Corresponding author for this work
  • CAS - Institute of Automation
  • University of Chinese Academy of Sciences
  • The University of Sydney
  • Chinese Academy of Sciences
  • Chinese Academy of Sciences

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Skeleton-based action recognition has drawn much attention recently. Previous methods mainly focus on using RNNs or CNNs to process skeletons. But they ignore the topological structure of the skeleton which is very important for action recognition. Recently, Graph Convolutional Networks (GCNs) achieve remarkable performance in modeling non-Euclidean structures. However, current graph convolutional networks lack the capacity of modeling hierarchical information, which may be sub-optimal for classifying actions which are performed in a hierarchical way. In this work, a novel Hierarchical Graph Convolutional Network (HiGCN) is proposed to deal with these problems. The proposed model includes several Hierarchical Graph Convolutional Layers (HiGCLs). Each layer consists of an attention block and a hierarchical graph convolutional block, which are used for salient feature enhancement and hierarchical representation learning, respectively. To represent hierarchical information of human actions, we propose a graph pooling method, which is differentiable and can be plugged into GCN in an end-to-end manner. Extensive experiments on two benchmark datasets show the state-of-the-art performance of our method.

Original languageEnglish
Title of host publicationImage and Graphics - 10th International Conference, ICIG 2019, Proceedings, Part 1
EditorsYao Zhao, Chunyu Lin, Nick Barnes, Baoquan Chen, Rüdiger Westermann, Xiangwei Kong
PublisherSpringer
Pages93-102
Number of pages10
ISBN (Print)9783030341190
DOIs
StatePublished - 2019
Externally publishedYes
Event10th International Conference on Image and Graphics, ICIG 2019 - Beijing, China
Duration: 23 Aug 201925 Aug 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11901 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference10th International Conference on Image and Graphics, ICIG 2019
Country/TerritoryChina
CityBeijing
Period23/08/1925/08/19

Keywords

  • Action recognition
  • Hierarchical graph convolutional network
  • Skeleton

Fingerprint

Dive into the research topics of 'Hierarchical Graph Convolutional Network for Skeleton-Based Action Recognition'. Together they form a unique fingerprint.

Cite this