A Lightweight Remote Sensing Aircraft Target Detection Network Based on Improved YOLOv5n
文献类型:会议论文
作者 | Wang, Jiale1,2; Bai, Zhe2![]() ![]() |
出版日期 | 2023 |
会议日期 | 2023-12-08 |
会议地点 | Hybrid, Chengdu, China |
关键词 | Deep learning lightweight network YOLOv5n ShuffleNetV2 Coordinate Attention EIoU loss |
DOI | 10.1109/ICCC59590.2023.10507627 |
页码 | 1678-1683 |
英文摘要 | Aiming at the problem that the target detection algorithm based on deep learning has a large number of network parameters and high requirements for computing resources, so that it is difficult to deploy on small hardware terminals. This paper proposes a lightweight remote sensing aircraft target detection network based on improved YOLOv5n, which fuses ShuffleNetV2 and Coordinate Attention (CA) mechanism. Firstly, we replace the Backbone part of the YOLOv5n network with the lightweight network ShuffleNetV2, which works well on target extraction and greatly reduces the model volume of the network; Secondly, we add the CA mechanism of cross-channel information to focus on direction perception and position information, and significantly improve the efficiency of target detection; Finally, in the loss function part, we use EIoU loss to speed up convergence and improve regression accuracy. The experimental results show that on the public dataset MAR20, our lightweight network extremely reduces the parameters by 84.4%, improves FPS by 6.5%, and reduces mAP@0.5 by 4.2% compared with the original YOLOv5n model. © 2023 IEEE. |
产权排序 | 1 |
会议录 | 2023 9th International Conference on Computer and Communications, ICCC 2023
![]() |
会议录出版者 | Institute of Electrical and Electronics Engineers Inc. |
语种 | 英语 |
ISBN号 | 9798350317251 |
源URL | [http://ir.opt.ac.cn/handle/181661/97476] ![]() |
专题 | 西安光学精密机械研究所_空间光学应用研究室 |
作者单位 | 1.University of Chinese Academy of Sciences, Beijing; 100049, China 2.Xi'an Institute of Optics and Precision Mechanics of Cas, Space Optics Technology Research Laboratory, Xi'an; 710119, China; |
推荐引用方式 GB/T 7714 | Wang, Jiale,Bai, Zhe,Zhang, Ximing,et al. A Lightweight Remote Sensing Aircraft Target Detection Network Based on Improved YOLOv5n[C]. 见:. Hybrid, Chengdu, China. 2023-12-08. |
入库方式: OAI收割
来源:西安光学精密机械研究所
浏览0
下载0
收藏0
其他版本
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。