基于YOLO框架的轻量化SAR图像舰船检测方法研究OA

To solve the problems of low accuracy and slow speed of ship detection in Synthetic Aperture Radar(SAR)images by existing target detection algorithms,a lightweight SAR ship detection algorithm based on YOLO framework is proposed.Firstly,based on the YOLO framework,the Ghost module and Efficient Channel Attention(ECA)mechanism are used to improve the ShuffleNetV2 network to construct a new backbone network,which reduces memory access costs and improves feature extraction capabilities.Secondly,the C3 module in the neck network is introduced into the multi-scale Pyramid Split Attention(PSA)module,which fully extracts spatial information from different scale feature maps and enhances the ability of multi-scale feature fusion.Finally,the lightweight GSConv convolution is used to eliminate the redundant features of the model,which reduces the number of model parameters while maintaining the detection accuracy.The experimental results show that on the public dataset SSDD,the average accuracy of the proposed model reaches 94.8％,the parameter number is 3.10 M,and the model weight size is 6.4 MB,which meets the real-time ship detection requirements of SAR image.