Human behaviour analysis based on spatio-temporal dual-stream heterogeneous convolutional neural network
by Qing Ye; Yuqi Zhao; Haoxin Zhong
International Journal of Computational Science and Engineering (IJCSE), Vol. 26, No. 6, 2023

Abstract: At present, there are still many problems to be solved in human behaviour analysis, such as insufficient utilisation of behaviour characteristic information and slow operation rate. We propose a human behaviour analysis algorithm based on spatio-temporal dual-stream heterogeneous convolutional neural network (STDNet). The algorithm is improved on the basic structure of the traditional dual-stream network. When extracting spatial information, the DenseNet uses a hierarchical connection method to construct a dense network to extract the spatial feature of the video RGB image. When extracting motion information, BNInception is used to extract temporal features of video optical flow images. Finally, feature fusion is carried out by multi-layer perceptron and sent to Softmax classifier for classification. Experimental results on the UCF101 dataset show that the algorithm can effectively use the spatio-temporal feature information in video, reduce the amount of calculation of the network model, and greatly improve the ability to distinguish similar actions.

Online publication date: Mon, 04-Dec-2023

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Computational Science and Engineering (IJCSE):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com