Title: Recognition method of basketball players' shooting action based on graph convolution neural network

Authors: Jin Xu

Addresses: School of Sports, Hunan City University, Yiyang, 413000, Hunan, China

Abstract: In order to improve the evaluation standard rate and accuracy of basketball players' shooting action recognition, a basketball players' shooting action recognition method based on graph convolution neural network is proposed. Using the learning method of three-dimensional graph convolution neural network, the optical flow data of collected video is analysed, the graph convolution neural network model is constructed, and the shooting action characteristics are extracted. The dynamic parameter identification of shooting action is realised by using graph neural network. The experimental test shows that the evaluation standard rate of shooting action recognition proposed in this paper is 0.8648, the accuracy of shooting action recognition is high, and the error is only 0.0066, which shows that the design method in this paper has good superior performance in improving the performance of basketball action recognition. It can provide technical support for the construction of basketball players' shooting action database.

Keywords: graph convolution neural network; basketball player; shooting; action recognition; skeleton sequence.

DOI: 10.1504/IJRIS.2022.126650

International Journal of Reasoning-based Intelligent Systems, 2022 Vol.14 No.4, pp.227 - 232

Received: 16 Feb 2022
Accepted: 20 May 2022

Published online: 31 Oct 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article