当前位置>主页 > 期刊在线 > 计算机技术 >

计算机技术2020年22期

基于卷积神经网络的视觉目标定位研究
温剑锋,覃一海
(广西建设职业技术学院 信息工程系,广西 南宁 530007)

摘  要:视觉目标定位是计算机视觉研究的重要方向之一,准确度高、稳定性好、速度快是视觉目标定位算法追求的目标,针对当前基于卷积神经网络的视觉目标定位算法在训练模型时需要持续训练和更新,导致计算量非常大、定位精度低、成功率低等不足,提出基于双卷积通道的卷积神经网络模型,通过与目前主流的定位算法进行比较,结果表明该算法具有较高的定位精度和成功率。


关键词:卷积神经网络;目标定位;图像检测



中图分类号:TP391.41;TP183         文献标识码:A          文章编号:2096-4706(2020)22-0113-03


Research on Visual Target Location Based on Convolution Neural Network

WEN Jianfeng,QIN Yihai

(Department of Information Engineering,Guangxi Polytechnic of Construction,Nanning 530007,China)

Abstract:Visual target localization is one of the important directions of computer vision research. High accuracy,good stability and high speed are the goals of visual target localization algorithm. In view of the shortcomings of the current visual target localization algorithm based on convolution neural network,which needs continuous training and updating in training model,resulting in large amount of calculation,low positioning accuracy and low success rate. In this paper,a convolution neural network model based on double convolution channels is proposed. Compared with the current mainstream positioning algorithms,the results show that the algorithm has higher positioning accuracy and success rate.

Keywords:convolution neural network;target location;image detection

基金项目:2020 年广西高校中青年教师科研基础能力提升项目:基于人工智能的视觉目标定位算法及并行技术研究(2020KY35012);2020 年广西建设职业技术学院校级教科研项目:基于深度学习的行人再识别研究(2020YB047)


参考文献:

[1] 张纯纯,冯创意,高统林. 基于机器视觉的目标识别与抓取研究 [J]. 农业装备与车辆工程,2019,57(12):93-96.

[2] 陈凯. 基于深度学习和回归模型的视觉目标跟踪算法研究 [D]. 武汉:华中科技大学,2018.

[3] 尹宽. 复杂情况下目标跟踪算法研究 [D]. 成都:四川师范大学,2020.

[4] DANELLJAN M,BHAT G,KHAN F S,et al. ECO:Efficient Convolution Operators for Tracking [C]//2017 IEEE Conferenceon Computer Vision and Pattern Recognition(CVPR).Honolulu:IEEE,2017:6931-6939.

[5] WANG N,ZHOU W G,TIAN Q,et al. Multi-CueCorrelation Filters for Robust Visual Tracking [C]//2018 IEEE/CVFConference on Computer Vision and Pattern Recognition(CVPR).SaltLake City:IEEE,2018:4844-4853.

[6] LI F,TIAN C,ZUO W M,et al. Learning Spatial-TemporalRegularized Correlation Filters for Visual Tracking [C]//2018 IEEE/CVFConference on Computer Vision and Pattern Recognition(CVPR).SaltLake City:IEEE,2018:4904-4913.


作者简介:

温剑锋(1979.12—),男,汉族,广西贵港人,系主任,副教授,硕士,研究方向:人工智能;

覃一海(1986.10—),男,汉族,广西北流人,专任教师,讲师,硕士,研究方向:人工智能。