• 首页
  • 期刊简介
  • 编委会
  • 投稿指南
  • 收录情况
  • 杂志订阅
  • 联系我们
引用本文:刘博文,步 扬,邹多宏,李建郎.一种基于残差网络改进的牙齿颜色分类模型[J].软件工程,2024,27(3):52-57.【点击复制】
【打印本页】   【下载PDF全文】   【查看/发表评论】  【下载PDF阅读器】  
←前一篇|后一篇→ 过刊浏览
分享到: 微信 更多
一种基于残差网络改进的牙齿颜色分类模型
刘博文1, 步 扬2, 邹多宏3, 李建郎1
(1.上海理工大学光电信息与计算机工程学院, 上海 200093;
2.中国科学院上海光学精密机械研究所, 上海 201899;
3.上海交通大学医学院附属第九人民医院, 上海 200011)
2385836330@qq.com; buyang@siom.ac.cn; zouduohongyy@163.com; lijianlang@usst.edu.cn
摘 要: 针对传统牙齿比色方法准确率低和效率低等问题,提出一种基于残差网络改进的牙齿颜色分类模型。该模型通过融合多层卷积结果以及引入压缩与激励注意力机制模块的方式,使网络能学习到更多的图像颜色特征。基于典型牙齿所建数据集进行颜色分类实验,在该数据集上对文中模型与GoogleNet、MobileNet-V1、ResNet-34和ResNet-50等模型进行颜色分类预测结果比较。实验结果表明,文中模型优于传统模型,预测分类准确度达到91.16%,有效提高了牙齿颜色分类准确率和效率。
关键词: 牙齿比色;颜色分类;深度学习;ResNet-18网络
中图分类号: TP751.2    文献标识码: A
基金项目: 国家重点研发计划(2020YFB2007504);上海市地方高校能力建设项目(22010503200);国家自然科学基金面上项目(61975217)
A Tooth Color Classification Model Improved by Residual Network
LIU Bowen1, BU Yang2, ZOU Duohong3, LI Jianlang1
(1.School of Optical-Electrical and Computer Engineering, University of Shanghai f or Science and Technology, Shanghai 200093, China;
2.Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201899, China;
3.Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200011, China)
2385836330@qq.com; buyang@siom.ac.cn; zouduohongyy@163.com; lijianlang@usst.edu.cn
Abstract: This paper proposes an improved tooth color classification model based on residual network to increase the accuracy and efficiency of traditional tooth colorimetric methods. This model enables the network to learn more image color features by fusing multilayer convolutional results and by introducing Squeeze-and-Excitation ( SE) attention mechanism module. Color classification experiments are conducted on a typical teeth dataset, on which the color classification prediction results of the proposed model are compared with those of GoogleNet, MobileNet-V1, ResNet-34 and ResNet-50. The experimental results show that the proposed model is better than the traditional models, and the prediction classification accuracy reaches 91.16% , which effectively improves the accuracy and efficiency of tooth color classification.
Keywords: tooth colorimetry; color classification; deep learning; ResNet-18 network


版权所有:软件工程杂志社
地址:辽宁省沈阳市浑南区新秀街2号 邮政编码:110179
电话:0411-84767887 传真:0411-84835089 Email:semagazine@neusoft.edu.cn
备案号:辽ICP备17007376号-1
技术支持:北京勤云科技发展有限公司

用微信扫一扫

用微信扫一扫