基于卷积神经网络的航道边坡病害识别  

Identification of channel slope diseases based on convolution neural network

作　　者：吴炼 钱国明[1] 韩晓健[2] WU Lian;QIAN Guoming;HAN Xiaojian(College of Electronic and Optical Engineering&College of Flexible Electronics,Nanjing University of Posts and Telecommunications,Nanjing 210023,China;College of Civil Engineering,Nanjing Tech University,Nanjing 211800,China)

机构地区：[1]南京邮电大学电子与光学工程学院、柔性电子学院,江苏南京210023 [2]南京工业大学土木工程学院,江苏南京211800

出　　处：《电子设计工程》2023年第15期90-93,共4页Electronic Design Engineering

摘　　要：内河航运是运输系统的关键组成部分,航道安全关系着水运系统的正常运转以及周边地区人民群众的安全。卷积神经网络技术在目标识别方面优异的效果为航道边坡病害检测提供了新的思路。针对人工检测效率低下的问题,文中设计了一种航道边坡病害分类识别系统,采集了三种典型航道边坡病害的图像数据集并进行了数据增强,使用改进的ResNet18网络进行训练,结果表明,该模型对边坡病害图像分类识别的精确率和召回率分别达到91.6%、91.53%,可有效替代人工检测方式。Inland shipping is a key part of the transportation system.The safety of the channel is connected with the normal operation of the water transportation system and the safety of the people in the surrounding areas.The excellent effect of convolutional neural network technology in target recognition provides new ideas for the detection of waterway slope diseases.For the problem of low efficiency of manual detection,this paper designs a channel slope disease classification and recognition system,collects the image data sets of three typical channel slope diseases,enhances the data,and uses the improved ResNet18 network for training.The results show that the precision and recall of the model for slope disease image classification and recognition are 91.6% and 91.53% respectively,which can effectively replace the manual detection method.

关 键 词：计算机应用 航道边坡病害识别 深度学习 深度残差网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]