基于遥感图像和改进YOLOv5s的舰船识别技术
Ship Recognition Technology Based on Remote Sensing Images and Improved YOLOv5s
张亮1
作者信息
- 1. 中国船舶集团有限公司第七二三研究所,江苏扬州 225101
- 折叠
摘要
航空航天与遥感技术的快速发展,带来了海量的高分辨率海面舰船图像信息,由于遥感图像带有复杂的背景环境,经典的YOLOv5s算法对舰船目标的检测与识别效果并不理想.针对这一问题,提出了一种改进的YOLOv5s模型,首先,利用Copy-paste进行数据增强.其次,构建C3Ghost模块替换原YOLOv5s中的C3模块,以减少网络的计算量.实验结果表明,改进的YOLOv5s模型对舰船目标检测与识别的效果有了显著提升.
Abstract
The rapid development of aerospace and remote sensing technology has brought a massive amount of high-resolution sea surface ship image information.Due to the complex background en-vironment of remote sensing images,the classical YOLOv5s algorithm is not ideal for detecting and recognizing ship targets.In response to this issue,an improved YOLOv5s model is proposed.First-ly,Copy paste is used for data augmentation.Secondly,a C3Ghost module is constructed to replace the C3 module in the original YOLOv5s to reduce the computation amount of the network.The ex-perimental results show that the improved YOLOv5s model has significantly improved the effec-tiveness of ship target detection and recognition.
关键词
遥感图像/YOLOv5s/舰船目标/检测与识别Key words
remote sensing image/YOLOv5s/ship target/detection and recognition引用本文复制引用
出版年
2024
NETL
NSTL
万方数据