基于改进YOLOv3-Tiny的遥感舰船小目标检测方法  

Small target detection of remote sensing ship based on improved YOLOv3-Tiny algorithm

作　　者：王欣 江涛[1] 马珍 魏玉梅 Wang Xin;Jiang Tao;Ma Zhen;Wei Yumei(School of Mathematics and Computer Science,Yunnan Minzu University,Kunming,Yunnan 650500,China)

机构地区：[1]云南民族大学数学与计算机科学学院,云南昆明650500

出　　处：《计算机时代》2023年第3期111-115,共5页Computer Era

基　　金：国家自然科学基金资助项目(61363022)。

摘　　要：遥感图像中各类舰船小目标检测存在检测难度大、检测精确度低等问题,因此提出一种基于改进YOLOv3-Tiny的遥感舰船小目标检测方法。先对训练数据进行Mosaic数据增强,丰富小尺度样本数据集,解决遥感图像小目标检测泛化能力低的问题,再在特征提取网络中加入混合域注意力机制CBAM,强化小目标的特征提取,提高对遥感舰船小目标的检测能力。实验结果表明,改进后的方法能够更加准确、快速地对遥感舰船小目标进行检测,显著提高了检测的准确率和召回率。There are many problems in the detection of small targets of ships in remote sensing images, such as high detection difficulty and low detection accuracy. Therefore, a remote sensing ship small target detection method based on improved YOLOv3-Tiny is proposed. Firstly, Mosaic data enhancement is carried out on the training data to enrich the small-scale sample data set and solve the problem of low generalization ability of small target detection in remote sensing images. Then, the hybrid domain attention module CBAM is added to the feature extraction network to strengthen the feature extraction of small targets and improve the detection ability of small targets in remote sensing ships. The experimental results show that the improved method can detect small targets of remote sensing ships more accurately and quickly, and significantly improve the detection accuracy and recall rate.

关 键 词：遥感图像 YOLOv3-Tiny 小目标检测 CBAM 数据增强 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]