改进YOLOv8的恶劣天气下船舶目标检测算法研究

打开文本图片集

关键词： 船舶检测； YOLOv8 算法； 恶劣天气； 聚核初始网络； 上下文锚点注意力模块； 特征识别中图分类号：TN911.73⁃34；TP391.4 文献标识码： A 文章编号：1004⁃373X（2025）12⁃0077⁃06

Abstract：In allusion to the difficulty of ship detection due to bad weather such as rain， snow and fog in existing inland waterways， a method of YOLOv8⁃based ship target detection for waterways， YOLOv8⁃Ship， is proposed. In this algorithm， a feature ⁃focused diffusion pyramid network integrating ADown module in YOLOv9 is proposed to make the features of each scale retain more context information. The poly kernel inception network （PKINet） and the context anchor attention （CAA） module are introduced to enhance C2f， thereby improving the features of the center region. The deep separable convolution is used to replace the conventional convolution in the backbone network， reducing the number of model parameters and computational complexity. The experimental results demonstrate that under weather conditions of rain， snow， and fog， in comparison with the traditional YOLOv8n， the accuracy of the improved algorithm is increased by 0.5% ， the recall rate is increased by 3.4% ， the F1 score is increased by 2% ， and the mAP @0.5 is increased by 1.2% . The average accuracy can reach 97.5% ， which effectively improves the recognition accuracy of passing ships in bad weather in inland waterways and has strong robustness.

Keywords：ship inspection; YOLOv8 algorithm; bad weather; PKINet; context anchor attention module; feature recognition

0 引 言

内河航运是综合运输体系以及水资源综合利用极为重要的组成部分，在多个方面都发挥着至关重要的作用，比如在有力地促进流域经济不断向前发展，有效地优化产业的布局规划，以及良好地服务于对外开放等方面，都展现出了独特而显著的意义。（剩余8575字）