现代电子技术2024，Vol.47Issue(7)：17-24,8.DOI:10.16652/j.issn.1004-373x.2024.07.003
基于弱光环境的车辆识别研究
Research on vehicle detection based on low light environments
摘要
Abstract
Since low light environment images show the problems of low exposure,foreground background fusion,and low contrast,it is difficult to effectively detect the vehicles(the objects)in the image in real time.Usually,it is required to design more complex neural network structures or establish additional comparative datasets to improve the image detection effects.However,these methods not only reduce the speed of the network,but also increase the training cost of the network.Therefore,a vehicle recognition network in low light environments is proposed.On the one hand,a feature extraction module and a feature fusion module are designed to improve the network's detection ability in low light environments.On the other hand,a module structure reparameterization method is used to increase the detection speed of the network.Experimental results show that the network can effectively identify vehicles in low light environment images while ensuring detection speed.关键词
弱光环境/车辆检测/神经网络/特征提取/特征融合/模块结构重参数化Key words
low light environment/vehicle detection/neural network/feature extraction/feature fusion/module structure reparameterization分类
电子信息工程引用本文复制引用
张峻祎,丁冰,丁洁..基于弱光环境的车辆识别研究[J].现代电子技术,2024,47(7):17-24,8.基金项目
国家重点研发计划(2022YFB3204600) （2022YFB3204600）
北京理工大学青年教师学术启动计划 （）
