轻量化的YOLOv4目标检测算法

doi:10.19678/j.issn.1000-3428.0062216

计算机工程 ›› 2022, Vol. 48 ›› Issue (8): 206-214. doi: 10.19678/j.issn.1000-3428.0062216

轻量化的YOLOv4目标检测算法

张宝朋, 康谦泽, 李佳萌, 郭俊宇, 陈少华

大连交通大学计算机与通信工程学院, 辽宁大连 116028

收稿日期:2021-07-30 修回日期:2021-09-30 发布日期:2021-10-11
作者简介:张宝朋(1995-),男,硕士研究生,主研方向为FPGA开发、车辆信息控制、目标检测;康谦泽、李佳萌、郭俊宇,硕士研究生;陈少华,教授、博士。
基金资助:
辽宁省自然科学基金面上项目“高速列车无线健康管理通信系统关键技术研究”（2021-MS-298）；辽宁省教育厅科学研究项目“列控系统故障诊断和预警机制的研究”（JDL2020006）。

Lightweight YOLOv4 Target Detection Algorithm

ZHANG Baopeng, KANG Qianze, LI Jiameng, GUO Junyu, CHEN Shaohua

School of Computer and Communication Engineering, Dalian Jiaotong University, Dalian, Liaoning 116028, China

Received:2021-07-30 Revised:2021-09-30 Published:2021-10-11

摘要/Abstract

摘要： YOLOv4目标检测算法主干网络庞大且参数量和计算量过多，难以部署在算力和存储资源有限的移动端嵌入式设备上。提出一种改进的YOLOv4目标检测算法，使用轻量化的ShuffleNet V2网络作为主干特征提取网络，更换模型激活函数及扩大卷积核，同时将YOLOv4网络中的普通卷积替换为深度可分离卷积，降低算法参数量、计算量和模型占用空间。在ShuffleNet V2网络结构的改进过程中分析并剪裁其基本组件，利用2个3 × 3卷积核级联的方式增强网络感受野，并使用Mish激活函数进一步提升网络检测精度和模型推理速度。在GPU平台和VisDrone 2020数据集上的实验结果表明，与YOLOv4算法相比，改进的YOLOv4算法在牺牲1.8个百分点的检测精度情况下，提高了27%的检测速度，压缩了23.7%的模型容量，并且能够充分发挥ZYNQ平台并行高速数据处理及低功耗的优势。

关键词: YOLOv4目标检测, ShuffleNet V2网络模型, 卷积运算, 轻量化网络, ZYNQ平台

Abstract: The YOLOv4 algorithm has a large backbone network and involves a large amount of parameters, due to which it is difficult to use the algorithmfor mobile embedded devices with limited computing power and storage.Aiming at solving these disadvantages, this study proposes an improved YOLOv4 target detection algorithm.It uses the lightweight ShuffleNetV2 network as the backbone feature extraction network, replaces the model activation function, and expands the convolution kernel.Moreover, the ordinary convolutions in the YOLOv4 network are replaced with depthwise separable convolutions, which reduce the amount of algorithm parameters and computations, and the size of the network model.Using the improved ShuffleNetV2 network structure, the basic components of the network structure are analyzed and clipped, the network receptive field is improved by cascading two 3×3 convolution cores, and the Mish activation function is used to further improve the network accuracy and speed of model reasoning.The experimental results on the GPU platform and the VisDrone 2020 dataset show that compared with the YOLOv4 algorithm, the proposed YOLOv4 algorithm improves the detection speed by 27% and compresses the algorithm capacity by 23.7%;however, detection accuracy is reduced by 1.8 percentage points.At the same time, the proposed algorithm can give full play to the advantages of parallel high-speed data processing and low power consumption of the ZYNQ platform.

Key words: YOLOv4 target detection, ShuffleNet V2 network model, convolution operation, lightweight network, ZYNQ platform

中图分类号:

TP393

张宝朋, 康谦泽, 李佳萌, 郭俊宇, 陈少华. 轻量化的YOLOv4目标检测算法[J]. 计算机工程, 2022, 48(8): 206-214.

ZHANG Baopeng, KANG Qianze, LI Jiameng, GUO Junyu, CHEN Shaohua. Lightweight YOLOv4 Target Detection Algorithm[J]. Computer Engineering, 2022, 48(8): 206-214.

http://www.ecice06.com/CN/Y2022/V48/I8/206

图/表 19

20220825092827

20220825092831

20220825092835

20220825092838

20220825092842

20220825092847

20220825092851

20220825092854

20220825092858

20220825092902

20220825092907

20220825092910

20220825092914

20220825092918

20220825092922

20220825092927

20220825092931

20220825092935

20220825092939

参考文献

[1] MEKHALFI M L, NICOLÒC, BAZI Y, et al.Detecting crop circles in Google earth images with mask R-CNN and YOLOv3[J].Applied Sciences, 2021, 11(5):2238.
[2] REN S Q, HE K M, GIRSHICK R, et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149.
[3] GIRSHICK R.Fast R-CNN[C]//Proceedings of IEEE International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2015:1440-1448.
[4] GAO X W, LI S Q, JIN B Y, et al.Intelligent crack damage detection system in shield tunnel using combination of RetinaNET and optimal adaptive selection[J].Journal of Intelligent &Fuzzy Systems, 2021, 40(3):4453-4469.
[5] REDMON J, DIVVALA S, GIRSHICK R, et al.You only look once:unified, real-time object detection[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2016:779-788.
[6] REDMON J, FARHADI A.YOLO9000:better, faster, stronger[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2017:6517-6525.
[7] LIU W, ANGUELOV D, ERHAN D, et al.SSD:single shot multibox detector[C]//Proceedings of European Conference on Computer Vision.Berlin, Germany:Springer, 2016:21-37.
[8] ZHENG Z, QI H Y, ZHUANG L, et al.Automated rail surface crack analytics using deep data-driven models and transfer learning[J].Sustainable Cities and Society, 2021, 70:102898.
[9] BOCHKOVSKIY A, WANG C Y, LIAO H Y M.YOLOv4:optimal speed and accuracy of object detection[EB/OL].[2021-06-04].https://arxiv.org/abs/2004.10934.
[10] WANG C Y, MARK L H Y, WU Y H, et al.CSPNet:a new backbone that can enhance learning capability of CNN[C]//Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2020:12-25.
[11] HE K M, ZHANG X Y, REN S Q, et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(9):1904-1916.
[12] MA N N, ZHANG X Y, ZHENG H T, et al.ShuffleNet V2:practical guidelines for efficient CNN architecture design[C]//Proceedings of European Conference on Computer Vision.Berlin, Germany:Springer, 2018:1-8.
[13] SANDLER M, HOWARD A, ZHU M L, et al.MobileNet V2:inverted residuals and linear bottlenecks[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2018:4510-4520.
[14] SERRA T, YU X, KUMAR A, et al.Scaling up exact neural network compression by ReLU stability[EB/OL].[2021-06-04].https://arxiv.org/abs/2102.07804.
[15] LUO W J, LI Y J, URTASUN R, et al.Understanding the effective receptive field in deep convolutional neural networks[EB/OL].[2021-06-04].https://arxiv.org/abs/1701.04128v2.
[16] SZEGEDY C, VANHOUCKE V, IOFFE S, et al.Rethinking the inception architecture for computer vision[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2016:2818-2826.
[17] MISRA D.Mish:a self regularized non-monotonic activation function[EB/OL].[2021-06-04].https://arxiv.org/abs/1908.08681v2.
[18] CHOLLET F.Xception:deep learning with depthwise separable convolutions[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2017:1800-1807.
[19] HOWARD A G, ZHU M L, CHEN B, et al.MobileNets:efficient convolutional neural networks for mobile vision applications[EB/OL].[2021-06-04].https://arxiv.org/abs/1704.04861.
[20] 刘超, 张晓晖, 胡清平.图像超分辨率卷积神经网络加速算法[J].国防科技大学学报, 2019, 41(2):91-97. LIU C, ZHANG X H, HU Q P.Image super resolution convolution neural network acceleration algorithm[J].Journal of National University of Defense Technology, 2019, 41(2):91-97.(in Chinese)
[21] DU D W, WEN L Y, ZHU P F, et al.VisDrone-DET2020:the vision meets drone object detection in image challenge results[C]//Proceedings of 2019 IEEE/CV International Conference on Computer Vision Workshop.Washington D.C., USA:IEEE Press, 2020:692-712.
[22] 孔维刚, 李文婧, 王秋艳, 等.基于改进YOLOv4算法的轻量化网络设计与实现[J].计算机工程, 2022, 48(3):181-188. KONG W G, LI W J, WANG Q Y, et al.Design and implementation of lightweight network based on improved YOLOv4 algorithm[J].Computer Engineering, 2022, 48(3):181-188.(in Chinese)
[23] KASPER-EULAERS M, HAHN N, BERGER S, et al.Short communication:detecting heavy goods vehicles in rest areas in winter conditions using YOLOv5[J].Algorithms, 2021, 14(4):114.

选择文件类型/文献管理软件名称

选择包含的内容

轻量化的YOLOv4目标检测算法

Lightweight YOLOv4 Target Detection Algorithm

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 19

参考文献

相关文章 4

编辑推荐

Metrics

本文评价

[1]	曾雷鸣, 侯进, 陈子锐, 周浩然. 基于弱语义分割的轻量化交通标志检测网络[J]. 计算机工程, 2022, 48(9): 269-276,285.
[2]	张法正, 杨娟, 汪荣贵, 薛丽霞. 基于动态自适应层叠网络的轻量化图像超分辨率重建[J]. 计算机工程, 2022, 48(12): 196-202.
[3]	林杰, 陈春梅, 刘桂华, 祝礼佳. 室内服务机器人的实时场景分割算法[J]. 计算机工程, 2021, 47(7): 21-29.
[4]	娄迎曦, 袁文浩, 彭荣群. 基于准循环神经网络的语音增强方法[J]. 计算机工程, 2020, 46(4): 316-320.

模态框（Modal）标题

选择文件类型/文献管理软件名称

选择包含的内容

轻量化的YOLOv4目标检测算法

Lightweight YOLOv4 Target Detection Algorithm

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 19

参考文献

相关文章 4

编辑推荐

Metrics

本文评价