火力与指挥控制2025，Vol.50Issue(8)：21-30,10.DOI:10.3969/j.issn.1002-0640.2025.08.003
一种轻量化伪装单兵目标检测算法
A Lightweight Target Detection Algorithm for Camouflaged Soldiers
摘要
Abstract
To address the issues of large model parameters and slow inference speeds in existing models,this paper proposes a lightweight target detection algorithm for camouflaged soldiers.The backbone of the algorithm is designed based on the lightweight network HGNetv2,utilizing the SRepVGG module for multi-scale feature fusion.Finally,partial convolution and 1×1convolution are combined in the coupled detection head.Compared with the baseline model YOLOv8,the deep learning network proposed in this article reduces the parameter count by 35.4%and improves inference speed by 18.9%while ensuring detection accuracy.This makes it more suitable for operation on edge computing devices with limited computational resources.关键词
轻量化/伪装/目标检测/边缘计算/骨干网络/特征融合Key words
lightweight/camouflage/object detection/edge computing/backbone network/feature fusion分类
信息技术与安全科学引用本文复制引用
张麟华,李腾,赵爽,富丽贞..一种轻量化伪装单兵目标检测算法[J].火力与指挥控制,2025,50(8):21-30,10.基金项目
国家自然科学基金青年科学项目(61602427) （61602427）
山西省科技厅重点研发基金资助项目(201903D121171) （201903D121171）
