计量学报2025，Vol.46Issue(11)：1622-1630,9.DOI:10.3969/j.issn.1000-1158.2025.11.10
基于改进YOLOv8的遥感图像目标检测算法
Remote Sensing Image Target Detection Algorithm Based on Improved YOLOv8
摘要
Abstract
Aiming at the problems of low detection accuracy caused by complex background and variable scale of remote sensing images,an improved YOLOv8 based remote sensing image target detection algorithm is proposed.First,based on the YOLOv8 network framework,the global attention mechanism is introduced in the backbone network to make the model pay more attention to important areas and reduce the interference of useless background on the model.Secondly,a wise crossover ratio is adopted as a new bounding box regression method to reduce the influence of low-quality examples on the model and improve the bounding box regression level of the model.Finally,an efficient multi-scale attention module and a dynamic detection head module are introduced into the feature fusion network and the head network respectively to improve the scale adaptability of the model.The experimental results show that the proposed algorithm achieves high detection accuracy on both DIOR and RSOD,and the average accuracy with a threshold value of 0.5 reaches 72.0% and 94.6%respectively.Moreover,the improved algorithm has strong robustness and can still maintain good detection performance when the input data set changes.关键词
计算机视觉/目标检测/深度学习/遥感图像/YOLOv8/动态检测头Key words
computer vision/target detection/deep learning/remote sensing image/YOLOv8/dynamic head分类
通用工业技术引用本文复制引用
JIN Mei,WANG Hongfeng,ZHANG Liguo,ZHANG Qi,YUAN Yulin..基于改进YOLOv8的遥感图像目标检测算法[J].计量学报,2025,46(11):1622-1630,9.基金项目
国家重点研发计划(2020YFB1711001) （2020YFB1711001）
河北省军民融合产业发展专项(2018B190) （2018B190）
