基于深度学习技术的船舶管件识别系统  

Recognition System of Ship Pipe Fittings Based on Deep Learning Technology

作　　者：胡敏[1] Hu Min(China Shipbuilding NDRI Engineering Co.,Ltd.,Shanghai 200090,China)

机构地区：[1]中船第九设计研究院工程有限公司,上海200090

出　　处：《科学与信息化》2022年第9期67-69,共3页Technology and Information

摘　　要：本文提出了船舶管件钢印智能识别的解决方案,基于深度学习技术设计了钢印智能识别系统。本文首先采用OpenCV技术对钢印图像进行必要的预处理,在此基础上运用Faster RCNN、RNN、LSTM、Tesseract OCR技术,对船舶管件进行钢印识别。其在测试集上的识别准确率为98.7%。在现场使用后表明,该识别系统具有较高的响应速度和可靠性,综合检测成功率达97.1%。This article puts forward the solution of intelligent identification of ship pipe steel seal.The intelligent identification system of steel seal is designed based on deep learning technology.Firstly,OpenCV technology is used to preprocess the steel seal image.On this basis,Faster RCNN,RNN,LSTM and Tesseract OCR technology are used to identify the steel seal of ship pipe fittings.The recognition accuracy on the test set is 98.7%.The field application shows that the recognition system has high response speed and reliability,and the comprehensive detection success rate is 97.1%.

关 键 词：深度学习 数字识别 船舶制造管理 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]