深度学习中的单阶段车辆检测算法综述  

Review of One-Stage Vehicle Detection Algorithms Based on Deep Learning

作　　者：杨锦帆 王晓强[1] 林浩 李雷孝 杨艳艳 李科岑 高静[4] YANG Jinfan;WANG Xiaoqiang;LIN Hao;LI Leixiao;YANG Yanyan;LI Kecen;GAO Jing(College of Information Engineering,Inner Mongolia University of Technology,Hohhot 010080,China;College of Computer Science and Engineering,Tianjin University of Technology,Tianjin 300384,China;College of Data Science and Application,Inner Mongolia University of Technology,Hohhot 010080,China;College of Computer and Information Engineering,Inner Mongolia Agricultural University,Hohhot 010011,China)

机构地区：[1]内蒙古工业大学信息工程学院,呼和浩特010080 [2]天津理工大学计算机科学与工程学院,天津300384 [3]内蒙古工业大学数据科学与应用学院,呼和浩特010080 [4]内蒙古农业大学计算机与信息工程学院,呼和浩特010011

出　　处：《计算机工程与应用》2022年第7期55-67,共13页Computer Engineering and Applications

基　　金：内蒙古自治区关键技术攻关计划项目(2019GG273);内蒙古自治区科技成果转化专项资金(2020CG0073,2021CG0033);内蒙古自治区科技重大专项(2019ZD015,2019ZD016);内蒙古自治区科技计划项目(2020GG0104)。

摘　　要：随着深度学习的发展,基于深度学习的车辆检测算法性能不断被提升,在构建智能交通体系方面发挥重要作用。单阶段目标检测模型因其检测速度的优越性,被广泛应用于车辆实时检测。为了综合分析基于深度学习的单阶段车辆检测算法相关改进及应用,分别对比了各类常用单阶段车辆检测算法,列举其改进措施以及在车辆检测方面存在的问题;重点阐述了基于常见单阶段车辆检测算法针对现有问题采取的相关改进以及应用领域;简要介绍了车辆检测相关数据集,对现阶段车辆检测中亟待解决的问题与难点进行了分析,提出了车辆检测未来的研究方向。With the development of deep learning, the performance of vehicle detection algorithm based on deep learning is continuously improved and plays an important role in building intelligent transportation system. One-stage target detection model is widely used in real-time vehicle detection in video stream due to its advantages in detection speed. In order to analyze the relevant improvements and applications of one-stage vehicle detection algorithm based on deep learning comprehensively, various common one-stage vehicle detection algorithms are compared by listing their improvements and problems left to be solved in vehicle detection, and then emphatically describes the relevant improvements and application fields of vehicle algorithm based on common one-stage framework. Finally, the data sets of vehicle detection are introduced briefly, the problems and difficulties to be solved in vehicle detection at the present stage are analyzed, and the feature research direction is put forward.

关 键 词：深度学习 车辆检测 单阶段目标检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]