基于改进Cascade⁃RCNN的内河航标检测算法  

Inland Navigation Mark detection Algorithmbased on Improved Cascade RCNN

作　　者：倪汉杰 初秀民[1,2] 张彬鹏 蒋仲廉 NI Hanjie;CHU Xiumin;ZHANG Binpeng;JIANG Zhonglian(National Engineering Research Center for Water Transport Safety,Wuhan University of Technology,Wuhan 430063,China;Intelligent Transportation System Research Center,Wuhan University of Technology,Wuhan 430063,China;School of Transportation and Logistics Engineering,Wuhan University of Technology,Wuhan 430063,China)

机构地区：[1]武汉理工大学国家水运安全工程技术研究中心,湖北武汉430063 [2]武汉理工大学智能交通系统研究中心,湖北武汉430063 [3]武汉理工大学交通与物流工程学院,湖北武汉430063

出　　处：《中国航海》2022年第3期99-105,共7页Navigation of China

基　　金：国家重点研发计划专项(2018YFB1600400);国家自然科学基金项目(52071250、51709220);中央高校基本科研业务费资助(2018IVB078)。

摘　　要：针对内河复杂环境下船舶智能航行对航标检测识别与分类技术的要求,通过对经典Faster⁃RCNN模型特征提取网络、锚框机制、检测框抑制算法、损失函数等进行优化,采用ResNeXt、Soft⁃NMS、GIoU等结构改进了级联Fas⁃ter⁃RCNN模型———Cascade⁃RCNN网络。以长江中游武桥水道航标数据为例,开展了算法测试与验证。研究结果表明:基于改进Cascade⁃RCNN的目标检测算法综合性能最佳,平均精度均值约94.17%、用时190毫秒/帧。该算法能够有效适应内河航标目标较小、重叠、多样的特点,保持较高的精确度与召回率,可满足内河复杂通航场景下航标的检测精度与效率需求。The algorithm is developed to answer the request of intelligent navigation for technologies for detection,identification and classification of navigation marks.The feature extraction network,anchor box mechanism,detection window suppression algorithm and loss function of typical faster RCNN are improved and the structure of the cascade RCNN is modified by introduction of ResNext,soft⁃NMS and GIoU(generalized intersection over union).The algorithm is verified with the data of navigation Marks in Wuhan waterway.Experiments show that the algorithm achieved average accuracy of 94.17%with the process speed of 190 ms per frame.This algorithm is seen strong in dealing with overlapping small targets.

关 键 词：内河航标 目标检测 卷积神经网络 

分 类 号：U697.31[交通运输工程—港口、海岸及近海工程] TN911.73[交通运输工程—船舶与海洋工程]