现代电子技术2025，Vol.48Issue(5)：127-134,8.DOI:10.16652/j.issn.1004-373x.2025.05.020
融合渐进式去雨网络的军用车辆检测算法
Military vehicle detection algorithm combining progressive deraining network
摘要
Abstract
In view of the accuracy degradation when detecting military vehicles(the objects)in the scenarios of rainy days,a military vehicle detection method that fuses a progressive rain removal algorithm with a high-accuracy detector is proposed.An image deraining algorithm named HISPNet is designed.The algorithm includes a lightweight and efficient rain streak feature extraction module and a cross-subnet rain streak feature fusion module,which mitigates the loss of detailed features during the process of convolution while capturing the rain streak information.The SPPFCSPC(spatial pyramid pooling and fully-connected spatial pyramid convolution)module is introduced to improve the single-stage detector,which ensures the detector sensing field and improves the efficiency at the same time,and enhances the representation ability of the detection model.The experimental results of the self-built dataset show that,in comparison with the classical detection algorithm YOLOv7,the mAP@0.5 and mAP@0.5:0.95 of the proposed algorithm have been improved by 4.4% and 2.8%,respectively,in the scenarios of rainy days,and its detection speed is 21.05 f/s.It can be seen that the proposed algorithm can basically meet the real-time requirements of detection,and is of effectiveness and practicality.关键词
图像去雨/编码器-解码器架构/轻量级高效雨纹特征提取模块/跨子网雨纹特征融合模块/SPPFCSPC模块/军用车辆检测Key words
image deraining/encoder-decoder architecture/lightweight and efficient rain streak feature extraction module/cross-subnet rain streak feature fusion module/SPPFCSPC module/military vehicle detection分类
信息技术与安全科学引用本文复制引用
苏胜君,仝秋红,柴国庆,苏海东,王凯,胡待方..融合渐进式去雨网络的军用车辆检测算法[J].现代电子技术,2025,48(5):127-134,8.基金项目
国家重点研发计划(2022YFC3002602) （2022YFC3002602）
"两链"融合企业(院所)联合重点专项-工业领域(2022LL-JB-03) （院所）
