一种基于多神经网络的烟支缺陷分类与定位方法  

A Method for Classification and Location of Cigarette Defects Based on Multi-Neural Network

作　　者：王端生 管一弘[1] 杨雄飞 崔云月 罗亚桃 黄岗 WANG Duan-sheng;GUAN Yi-hong;YANG Xiong-fei;CUI Yun-yue;LUO Ya-tao;HUANG Gang(School of Science,Kunming University of Science and Technology,Yunnan 650500,China;Kunming Julin Technology Company Limited,Yunnan 650000,China)

机构地区：[1]昆明理工大学理学院,云南昆明650500 [2]昆明聚林科技有限公司,云南昆明650000

出　　处：《软件导刊》2022年第2期184-188,共5页Software Guide

摘　　要：为提高烟支生产过程中烟支缺陷检测效率及判断烟支缺陷产生的原因,提出基于多神经网络的烟支缺陷分类与定位方法。该方法首先对烟支数据集进行图像预处理,再采用改进VGG19网络对烟支是否存在缺陷进行预测,利用YOLOV3网络对缺陷烟支进行定位和类别判定,最后根据统计结果推断产生缺陷的原因,查找易产生缺陷的工艺步骤。研究表明,该方法检测精度高,缺陷定位和类别分类准确。In order to improve the detection efficiency of cigarette defects in the process of cigarette production and to find the reasons of cigarette defects,propose a method for classification and location of cigarette defects based on multi-neural networks. This method first performs image preprocessing on the cigarette data set,then uses the improved VGG19 network to predicting whether there are defects in the cigarettes,using the YOLOV3 network to locate and determine of the cigarettes which are predicted have defects in the cigarette data. Finally,according to the cigarette statistical results of the defect categories inversely infer the causes of the defects in the cigarettes and locate the wrong process in time,so as to realize the detection of cigarette defects. Research shows that the detection method has high detection accuracy,accurate defect category analysis and accurate defect positioning.

关 键 词：烟支缺陷 神经网络 VGG19 YOLOV3 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP391.41[自动化与计算机技术—控制科学与工程]