YOLOv7的实时遥感图像舰船目标旋转检测算法分析
Analysis of Real time Remote Sensing Image Ship Target Rotation Detection Algorithm for YOLOv7
周春云1
作者信息
- 1. 扬州万方科技股份有限公司,江苏 225000
- 折叠
摘要
阐述一种用于检测遥感图像中带有旋转角度舰船的YOLOv7算法,在YOLOv7的基础上加入角度分量信息检测舰船旋转角度,在特征提取部分加入NAM注意力模块以保证不损失速度的前提下提高对大尺度遥感图像中检测目标的准确度.探讨在异构AI计算环境中,通过自建遥感舰船图像数据集,完成训练和推理验证.
Abstract
This paper describes a YOLOv7 algorithm for detecting ships with rotation angles in remote sensing images.Based on YOLOv7,angle component information is added to detect the rotation angle of ships.In the feature extraction section,a NAM attention module is added to improve the accuracy of detecting targets in large-scale remote sensing images without losing speed.Explore training and inference validation through self built remote sensing ship image datasets in heterogeneous AI computing environments.
关键词
智能算法/YOLOv7/遥感图像/旋转目标检测Key words
intelligent algorithm/YOLOv7/remote sensing image/rotating object detection引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据