融合注意力及路径聚合机制的装甲车识别方法

丛玉华; 王志胜; 邢长达

doi:10.15892/j.cnki.djzdxb.2021.05.027

弹箭与制导学报 >

2021 , Vol. 41 >Issue 5: 138 - 144

DOI: https://doi.org/10.15892/j.cnki.djzdxb.2021.05.027

融合注意力及路径聚合机制的装甲车识别方法

丛玉华 ,
王志胜 ,
邢长达

展开

¹ 南京理工大学紫金学院,南京210023
² 南京航空航天大学自动化学院,南京 211106
³ 南京航空航天大学深圳研究院,广东深圳 518063

丛玉华(1981-),女,山东烟台人,博士研究生,研究方向:无人机集群规划与控制.

收稿日期: 2021-07-27

网络出版日期: 2025-05-30

基金资助

国家自然科学基金青年科学基金(62101247);深圳市中央引导地方科技发展专项资金(2021Szvup063)

收起

Armored Vehicle Recognition Method Integrating Attention and Path Aggregation Mechanism

CONG Yuhua ,
WANG Zhisheng ,
XING Changda

Expand

¹ Nanjing University of Science and Technology ZiJin College, Nanjing 210023, China
² College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China
³ Shenzhen Research Institute of Nanjing University of Aeronautics and Astronautics, Guangdong Shenzhen 518063, China

Received date: 2021-07-27

Online published: 2025-05-30

Fold

摘要

基于卷积神经网络的Yolov4-Tiny以其轻量化和高速性适用于战场环境下装甲车的识别,但是以损失精度换取速度,因此需要在保证一定识别速度基础上对其进行改进。首先在主干网络引入注意力机制,在通道和空间上加强关键特征提取;然后引入路径聚合方法,在特征金字塔结构基础上融入自下而上的路径聚合方法,加强不同尺度特征的提取。通过5种网络结构对装甲车识别的效果对比,改进后的网络在保证轻量级特性基础上,实时速度较快,精确度有大幅提升,证明改进的有效性。

关键词： 装甲车; 目标识别; Yolov4-Tiny; 注意力机制; 路径聚合机制

本文引用格式

丛玉华 , 王志胜 , 邢长达 . 融合注意力及路径聚合机制的装甲车识别方法[J]. 弹箭与制导学报, 2021 , 41(5) : 138 -144 . DOI: 10.15892/j.cnki.djzdxb.2021.05.027

Abstract

Yolov4Tiny based on convolutional neural network is suitable for the recognition of armored vehicles in battlefield environment because of its lightweight and high speed, but it exchanges speed with loss of accuracy. Therefore, it needs to be improved on the basis of ensuring a certain recognition speed. Firstly, the attention mechanism is introduced into the backbone network to strengthen the key feature extraction in channel and space. Then the path aggregation method is introduced, and the bottomup path aggregation method is integrated into the feature pyramid structure to strengthen the extraction of different scale features. Through the comparison of the effects of five network structures on armored vehicle recognition, the improved network has fast realtime speed and greatly improved accuracy on the basis of ensuring lightweight characteristics, which proves the effectiveness of the improvement.

Key words： armored car; target identification; Yolov4-Tiny; attention mechanism; path aggregation mechanism

参考文献

[1]	刘英 . 基于卷积神经网络的陆战场目标分类算法研究 [D]. 成都 : 电子科技大学 , 2020 .
[2]	程学生 , 姚旺生 . 一种坦克装甲车识别方法 [J]. 电子测量技术 , 2007,30 (12): 57 - 58 .
[3]	顾超越 , 李喆 , 史晋涛 , 等 . 基于改进 Faster-RCNN 的无人机巡检架空线路销钉缺陷检测 [J]. 高电压技术 , 2020(9): 3089 - 3096 .
[4]	REDMON J , DIVVALA S , GIRSHICK R , et al . You only look once: unified, real-time object detection [C]// Anon. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2016:779 - 788 .
[5]	REDMON J , FARHADIA A . YOLO9000: better, faster, stronger [C]// IEEE. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2017:7263 - 7271 .
[6]	REDMON J , FARHADI A . Yolov3: an incremental improvement [J]. ArXiv E-prints , 2018(4): 1 - 5 .
[7]	BOCHKOVSKIY A , WANG CY , LIAO HY . YOLOV4: Optimal speed and accuracy of object detection [J]. Computer Vision and Pattern Recognition , 2020,17 (9): 198 - 215 .
[8]	郭紫嫣 , 韩慧妍 , 何黎刚 , 等 . 基于改进的YOLOV4的手势识别算法及其应用 [J]. 中北大学学报 , 2021,42 (3): 223 - 231 .
[9]	马超杰 , 杨华 , 吴丹 , 等 . 自动目标识别技术在武器系统中的应用 [J]. 飞航导弹 , 2008(10): 45 - 48 .
[10]	CHUA L O , ROSKA T . The CNN paradigm [J]. IEEE Transactions on Circuits and Systems I-regular Papers , 1993,40 (3): 147 - 156 .
[11]	KRIZHEVSKY A , SUTSKEVER I , HINTON G E . Imagenet classification with deep convolutional neural networks [J]. Communications of the ACM , 2017,60 (6): 84 - 90 .
[12]	OQUAB M , BOTTOU L , LAPTEV I , et al . Learning and transferring mid-level image representations using convolutional neural networks [C]// IEEE. Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition , 2014:1717 - 1724 .
[13]	GOODFELLOW I , BENGIO Y , COURVILLE A . Deep learning [M]. Cambridge : MIT Press , 2016:326 - 366 .
[14]	BRUNA J , SZLAM A , LECUN Y . Signal recovery from pooling representations [J]. Statistics , 2014,32 . 307 - 315 .
[15]	WOO S , PARK J , LEE JY , et al . CBAM: Convolutional block attention module [C]// ECCV. Proceedings of the 2018 European Conference on Computer Vision , 2018:1 - 17 .
[16]	LIU S , QI L , QIN H , et al . Path aggregation network for instance segmentation [C]// IEEE. Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2018:1 - 11 .

Options

文章导航

模态框（Modal）标题

摘要

本文引用格式

Abstract

参考文献