现代信息科技2025，Vol.9Issue(10)：50-57,8.DOI:10.19850/j.cnki.2096-4706.2025.10.010
基于YOLOv8-cls改进算法的移网覆盖图片分类方法
Mobile Network Coverage Image Classification Method Based on Improved YOLOv8-cls Algorithm
唐天彪 1王舒波1
作者信息
- 1. 中国联合网络通信有限公司 重庆市分公司,重庆 401121
- 折叠
摘要
Abstract
The mobile network coverage problem is the research hotspot in the field of communication.The traditional MR.TadvRsrp image classification and diagnosis relies on the skills and experience of engineers,which is cumbersome and has poor real-time performance.In order to solve this problem,this paper proposes an intelligent classification method based on improved YOLOv8-cls algorithm.This method uses YOLOv8n-cls network as the feature extraction network,integrates the mixed attention module(CBAM)in the backbone network to improve the feature extraction ability,and adds the Spatial Pyramid Pooling-Fast(SPP-Fast)module to enhance the multi-scale feature aggregation effect.By designing a multi-level receptive field fusion structure(MF),the problem of insufficient feature fusion caused by the lack of neck structure(Neck)in the YOLOv8n-cls network is compensated.The improved algorithm significantly improves the accuracy and processing speed of image classification.The experimental results show that the classification and diagnosis accuracy of this method for 4G/5G network coverage slice image from specific areas reaches 95.7%,and the processing speed(FPS)reaches 110 frames/second.It exhibits high accuracy and strong real-time performance,offering efficient evaluation support for network planning and operational optimization.关键词
移网覆盖/图片分类/多尺度融合/YOLOv8-clsKey words
mobile network coverage/image classification/multi-scale fusion/YOLOv8-cls分类
信息技术与安全科学引用本文复制引用
唐天彪,王舒波..基于YOLOv8-cls改进算法的移网覆盖图片分类方法[J].现代信息科技,2025,9(10):50-57,8.
