深度学习在陶瓷表面缺陷检测方面的研究进展  

Progress in Deep Learning for Surface Defect Detection of Ceramics

作　　者：周曼 吴天钊 代宝鑫 许新统 孔令兵 梁立新 ZHOU Man;WU Tianzhao;DAI Baoxin;XU Xintong;KONG Lingbing;LIANG Lixin(College of New Materials and New Energies,Shenzhen University of Technology,Shenzhen 518118,Guangdong,China;College of Applied Technology,Shenzhen University,Shenzhen 518060,Guangdong,China;College of Integrated Circuits and Optoelectronic Chips,Shenzhen Technology University,Shenzhen 518118,Guangdong,China;College of Big Data and Internet,Shenzhen Technology University,Shenzhen 518118,Guangdong,China)

机构地区：[1]深圳技术大学新材料与新能源学院,广东深圳518118 [2]深圳大学应用技术学院,广东深圳518060 [3]深圳技术大学集成电路与光电芯片学院,广东深圳518118 [4]深圳技术大学,大数据与互联网学院,广东深圳518118

出　　处：《陶瓷学报》2023年第5期874-884,共11页Journal of Ceramics

基　　金：广东省基础与应用基础研究基金(2020B1515120002);深圳市高等院校稳定支持面上项目(SZWD2021001)。

摘　　要：针对陶瓷表面缺陷检测问题,深度学习算法是近年来研究的热点之一。通过建立合适的数据集、合适的网络模型和算法,可以实现对陶瓷表面缺陷的自动检测和分类。目前,常用的深度学习表面缺陷检测算法包括卷积神经网络(Convolutional Neural Network,CNN)、循环神经网络(Recurrent Neural Network,RNN)和多层感知器(Multilayer Perceptron,MLP)等。其中,基于YOLOv5算法的陶瓷缺陷检测方法是近期较为先进的一种方法,它具有较高的检测精度和实时性,能够准确地检测和识别陶瓷表面的各种缺陷,通过优化网络结构和损失函数,还可以进一步提高算法的性能;基于CSS算法的陶瓷缺陷检测方法提出使用图像分割的方法来分割陶瓷缺陷样本,并对分割后的样本集图像做二值化处理,突出缺陷的位置和大小。综述了陶瓷与深度学习相结合在材料表面缺陷检测方面的研究进展,并介绍了基于深度学习算法的陶瓷缺陷检测方法,以及详细综述了基于YOLOv5和基于CSS的陶瓷表面缺陷检测算法过程。Aiming at the problem of ceramic surface defect detection,deep learning algorithm is one of the hot spots in recent research.By establishing suitable data sets,selecting appropriate network models and algorithms,automatic detection and classification of ceramic surface defects can be realized.Commonly used deep learning surface defect detection algorithms include Convolutional Neural Network(CNN),Recurrent Neural Network(RNN),and Multilayer Perceptron(MLP),etc.Among them,the ceramic defect detection method based on YOLOv5 algorithm is a relatively advanced method in recent years,which has high detection accuracy and real-time performance,can accurately detect and identify various defects on the surface of ceramics and can further improve the performance of the algorithm by optimizing the network structure and loss function.The ceramic defect detection method based on CSS algorithm is to use the image segmentation method to segment ceramic defect samples and perform binary processing on the segmented sample set images to highlight the position and size of the defects.This paper was aimed to review the research progress in deep learning for surface defect detection of ceramics,introduce ceramic defect detection methods based on deep learning algorithms and summarize the process of ceramic surface defect detection algorithms based on YOLOv5 and CSS.

关 键 词：深度学习 材料科学与工程 机器视觉 陶瓷表面缺陷检测 

分 类 号：TQ174.75[化学工程—陶瓷工业]