舰船电子工程2026，Vol.46Issue(2)：50-54,194,6.DOI:10.3969/j.issn.1672-9730.2026.02.011
结合混合注意力机制的客滚船危险品检测研究
Research on Dangerous Goods Detection for Ro/Ro Passenger Ships Combined with Hybrid Attention Mechanism
摘要
Abstract
To address the issue of traditional instrument equipment and manual inspection being the main reliance for danger-ous goods detection on Ro/Ro passenger ships,a Faster RCNN algorithm combining hybrid attention mechanism is proposed.First-ly,the deep residual network ResNet50 is introduced to replace the Faster RCNN network's VGG16 for feature extraction.Then,a hybrid attention mechanism is introduced after the region generation network,aiming to mine spatiotemporal information and im-prove detection and classification performance.A large number of experimental results show that compared to existing object detec-tion algorithms,the proposed algorithm has better classification performance for dangerous goods detection,with an average classifi-cation result of 90.27%.关键词
客滚船/危险品检测/Faster RCNN/ResNet50/混合注意力机制Key words
Ro/Ro passenger ships/dangerous goods detection/Faster RCNN/ResNet50/hybrid attention mechanism分类
交通工程引用本文复制引用
姚竞争,李至立,张耀刚..结合混合注意力机制的客滚船危险品检测研究[J].舰船电子工程,2026,46(2):50-54,194,6.基金项目
山东省重点研发计划(重大科技创新工程)项目(编号:2021CXGC010702)资助. （重大科技创新工程）
