Graph convolutional network with STC attention and adaptive normalization for skeleton-based action recognition.
Saved in:
| Title: | Graph convolutional network with STC attention and adaptive normalization for skeleton-based action recognition. |
|---|---|
| Authors: | Zhou, Haiyun1 (AUTHOR), Xiang, Xuezhi2,3 (AUTHOR) xiangxuezhi@hrbeu.edu.cn, Qiu, Yujian2 (AUTHOR), Liu, Xuzhao2 (AUTHOR) |
| Source: | Imaging Science Journal. Oct2023, Vol. 71 Issue 7, p636-646. 11p. |
| Subjects: | Mathematical decoupling, Weight training, Skeleton |
| Abstract: | Graph Convolutional Network (GCN) have been widely used in the field of skeleton-based action recognition and have achieved exciting results. Introducing attention mechanism in the process of extracting skeleton features has always been a hot spot in GCN-related research. In this article, we design a new graph convolutional network, which combines the advanced decoupling graph convolutional network (DC-GCN) with spatial, temporal, channel (STC) series attention module and adaptive normalization (AN). The STC attention module helps the network tend to extract important information from skeleton features. In addition, in order to improve the adaptability of the normalization method to GCN, we design the AN module instead of the BN module, which can train the weights of different normalization methods, so that each normalization layer in the network adopts the most suitable normalization operation. The experimental results show that the accuracy of our method is competitive with the state-of-the-art action recognition methods. [ABSTRACT FROM AUTHOR] |
| Copyright of Imaging Science Journal is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | Graph Convolutional Network (GCN) have been widely used in the field of skeleton-based action recognition and have achieved exciting results. Introducing attention mechanism in the process of extracting skeleton features has always been a hot spot in GCN-related research. In this article, we design a new graph convolutional network, which combines the advanced decoupling graph convolutional network (DC-GCN) with spatial, temporal, channel (STC) series attention module and adaptive normalization (AN). The STC attention module helps the network tend to extract important information from skeleton features. In addition, in order to improve the adaptability of the normalization method to GCN, we design the AN module instead of the BN module, which can train the weights of different normalization methods, so that each normalization layer in the network adopts the most suitable normalization operation. The experimental results show that the accuracy of our method is competitive with the state-of-the-art action recognition methods. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 13682199 |
| DOI: | 10.1080/13682199.2023.2190927 |
