数字海洋与水下攻防2025，Vol.8Issue(5)：575-586,12.DOI:10.19838/j.issn.2096-5753.2025.05.007
一种跨级特征自适应融合的轻量化水声目标检测模型
Lightweight Underwater Acoustic Target Detection Model Based on Cross-Level Feature Adaptive Fusion
摘要
Abstract
To address the challenges of high noise,feature loss,and low resolution in sonar image target detection,an adaptive cross-stage lightweight underwater acoustic target detection model is proposed in this study.In terms of methods,a lightweight feature extraction network with multi-scale attention is designed to enhance feature focusing,while the conventional SPPF is replaced by a focus modulation network to improve localization and recognition of key regions.Furthermore,an improved adaptive spatial feature fusion module is introduced in the prediction stage to expand the receptive field and strengthen both physical and semantic representations while maintaining compactness.Experimental results on real sonar datasets show that compared with existing methods,the proposed model achieves fewer parameters,the detection accuracy is improved by approximately 4.7%,and the speed is increased by more than 18%.The findings indicate that the model achieves a favorable balance among accuracy,efficiency,and complexity,offering an effective approach for future designs of underwater acoustic target detection.关键词
海洋/声呐图像/特征增强/多级目标检测Key words
marine/sonar image/feature enhancement/multi-level target detection分类
信息技术与安全科学引用本文复制引用
ZHENG Kun,LU Ritian,ZHAO Yanghong,CHEN Zhe,XIE Guohao,QIU Hongbing..一种跨级特征自适应融合的轻量化水声目标检测模型[J].数字海洋与水下攻防,2025,8(5):575-586,12.基金项目
广西技术创新引导专项"基于双频多波束声呐的水下人造物高分辨探测研究"(桂科 AC25069006) （桂科 AC25069006）
广西科技基地和人才专项"基于深度学习的声呐图像识别方法研究"(桂科 AD21220098). （桂科 AD21220098）
