Virtual sports interactive system design integrating ghost net network and improved YOLOv5 algorithm

Abstract

With the development of virtual reality, the human–computer interaction through virtual sports is gradually maturing, and users are gradually looking to interact with the two-dimensional world. The research on this type of algorithm has gained attention. However, due to the delay of the old transmission technology in the transmission of pictures, which is higher than the reaction time of human brain, the pictures are inconsistent and illogical, and the user interaction experience is poor. To solve it, this research realizes the fusion of ghost network and You Only Look Once version 5, and the simulation experiment is carried out on the data set. Firstly, the convolution block attention module is inserted into the You Only Look Once version 5 algorithm to optimize its way of calculating Hadamard product. Then, the improved algorithm and the ghost network generation fusion algorithm are combined through the direct channel. Then the algorithm is combined with the virtual sports interactive system to upgrade its key point rearrangement mode. Finally, the performance of the system is characterized on Javelin dataset, and the stability is compared with the other three algorithms. The average score of the six experiments of the system is 9.5, and the average scores of You Only Look Once version 5, ghost network and particle swarm optimization algorithm are 9.42, 9.28 and 9.36, respectively. Results show that this model has excellent performance in adjusting data volatility, and is extensive in virtual sports interaction, which can effectively improve the user experience.

Keywords

• ghost net network
• yolov5 algorithm
• virtual reality technology
• sports interactive system