Please wait a minute...
浙江大学学报(理学版)  2021, Vol. 48 Issue (6): 676-683    DOI: 10.3785/j.issn.1008-9497.2021.06.005
数学与计算机科学     
基于改进的DenseNet-BC对少数民族服饰的识别
杨冰, 徐丹, 张豪远, 罗海妮
云南大学 信息学院, 云南 昆明 650500
Minority clothing recognition based on improved DenseNet-BC
YANG Bing, XU Dan, ZHANG Haoyuan, LUO Haini
School of Information,Yunnan University,Kunming 650500,China
 全文: PDF(1000 KB)   HTML  
摘要: 随着信息技术的发展,数字技术越来越多地应用于民族文化数字化保护,民族服饰的数字化及分类问题也日益受关注。相比一般服饰,少数民族服饰具有更多的细节特征信息,对其进行分类识别具有很大挑战。选用卷积神经网络DenseNet-BC作为基础网络结构,设计并使用了多尺度密集连接单元,用不同大小的卷积提取不同尺度的特征信息,以提高网络的学习能力;此外,为进一步提高网络的鲁棒性,提出一种局部和全局注意力机制方法进行分类识别。实验结果表明,改进的DenseNet-BC模型对少数民族服饰的识别准确率达95.18%,较ResNet-18、ResNet-34和DenseNet模型的识别准确率分别提升了3.84%、2.27%和1.18%。改进的DenseNet-BC模型具有更好的特征提取能力,能够提取更多的细节特征信息,一定程度上解决了普通模型提取特征尺度单一、特征丰富度低的问题。
关键词: 民族服饰分类图像识别DenseNet注意力机制多尺度    
Abstract: With the development of information technology,the digital protection of national culture has received more and more attention,the digitalization and classification of national costumes have also become a hot topic.Compared with general clothing,ethnic minority clothing holds more detailed feature information,which poses a big challenge to classify and identify them.This paper selects DenseNet-BC as the basic network structure,and uses multi-scale dense connection units to extract feature information at different scales through different convolution sizes hence improving the learning ability of the network.Furthermore,a local and global attention mechanism is proposed for classification and recognition to make the network robust.Experimental results show that the recognition accuracy of the improved DenseNet-BC model is 95.18%,which is respectively 3.84%,2.27%,and 1.18% higher than the recognition accuracy of ResNet-18,ResNet-34 and DenseNet models.The improved DenseNet-BC model has also better feature extraction capabilities,and can extract more detailed feature information.In brief,this network solves partially the problem of single feature scale and low feature richness of common models.
Key words: national costume classification    image recognition    DenseNet    attention mechanism    multi-scale
收稿日期: 2021-03-19 出版日期: 2021-11-25
CLC:  TP 391  
基金资助: 国家自然科学基金资助项目(61540062,61761046);云南省“万人计划”云岭学者专项;云南省科技厅—云南大学“双一流”建设联合基金项目(2019FY003012).
通讯作者: ORCID:https://orcid.org/0000-0003-4602-3550,E-mail:danxu@ynu.edu.cn.     E-mail: danxu@ynu.edu.cn
作者简介: 杨冰(1995—),ORCID:https://orcid.org/0000-0001-5412-5683,女,硕士,主要从事计算机图像处理和视觉研;
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章  
杨冰
徐丹
张豪远
罗海妮

引用本文:

杨冰, 徐丹, 张豪远, 罗海妮. 基于改进的DenseNet-BC对少数民族服饰的识别[J]. 浙江大学学报(理学版), 2021, 48(6): 676-683.

YANG Bing, XU Dan, ZHANG Haoyuan, LUO Haini. Minority clothing recognition based on improved DenseNet-BC. Journal of Zhejiang University (Science Edition), 2021, 48(6): 676-683.

链接本文:

https://www.zjujournals.com/sci/CN/10.3785/j.issn.1008-9497.2021.06.005        https://www.zjujournals.com/sci/CN/Y2021/V48/I6/676

1 国务院办公厅.关于实施中华优秀传统文化传承发展工程的意见[EB/OL].[2017-01-25].http://www.gov.cn/zhengce/2017-01/25/content_5163472.htm. doi:10.3969/j.issn.1006-9607.2017.02.018 General Office of the State Council.Opinions on the Implementation of the Inheritance and Development Project of Chinese Excellent Traditional Culture[EB/OL]. [2017-01-25]. http://www.gov.cn/zhengce/2017-01/25/content_5163472.htm. doi:10.3969/j.issn.1006-9607.2017.02.018
2 国务院办公厅.国务院关于公布第二批国家级非物质文化遗产名录和第一批国家级非物质文化遗产扩展项目名录的通知[EB/OL]. [2008-06-24]. http://www.gov.cn/zwgk/2008-06/14/content_1016331.htm. doi:10.1055/s-002-10110 General Office of the State Council. Notice of the State Council on Issuing the Second Batch of National Intangible Cultural Heritage List and the First Batch of National Intangible Cultural Heritage Expansion Project List[EB/OL]. [2008-06-24]. http://www.gov.cn/zwgk/2008-06/14/content_1016331.htm. doi:10.1055/s-002-10110
3 SHEN J,LIU G C,CHEN J,et al.Unified structured learning for simultaneous human pose estimationand garment attribute classification[J]. IEEE Transactions on Image Processing,2014,23(11):4786-4798. DOI:10.1109/TIP.2014.2358082
4 BOSSARD L,DANTONE M,LEISTNER C,et al.Apparel classification with style[C]//Proceedings of the 11th Asian Conference on Computer Vision.Berlin /Heidelberg:Springer ,2012:321-335. DOI:10.1007/978-3-642-37447-0_25
5 CHEN H Z,GALLAGHER A,GIROD B.Describing clothing by semantic attributes[C]// Proceedings of the 12th European Conference on Computer Vision.Florence:Springer,2012:609-623. DOI:10.1007/978-3-642-33712-3_44
6 SURAKARIN W,CHONGSTITVATANA P.Predicting types of clothing using SURF and LDP based on bag of features[C]// International Conference on Electrical Engineering/ Electronics Computer,Telecommunications and Information Technology.Hua Hin:IEEE,2015:1-5. DOI:10. 1109/ECTICon.2015.7207101
7 LAO B,JAGADEESH K.Convolutional Neural Networks for Fashion Classification and Object Detection[EB/OL].[2015-09-07].http://cs231n.stanford.edu/reports/2015/pdfs/BLAO_KJAG_CS231N_FinalPaperFashionClassification.pdf
8 DONG C Y,SHI Y Q,TAO R.Convolutional neural networks for clothing image style recognition[C]// 2018 International Conference on Computational,Modeling,Simulation and Mathematical Statistics.Pennsylvania:CMSMS,2018:592-597. doi:10.12783/dtcse/cmsms2018/25262
9 包青平,孙志锋.基于度量学习的服饰图像分类和检索[J].计算机应用与软件,2017,34(4):255-259. DOI:10.3969/j.issn.1000-386x.2017.04.043 BAO Q P,SUN Z F.Clothing image classification and retrieval based on metric learning[J].Computer Applications and Software,2017,34(4):255-259. DOI:10.3969/j.issn.1000-386x.2017.04.043
10 吴圣美,刘骊,付晓东,等.结合人体检测和多任务学习的少数民族服饰识别[J].中国图象图形学报,2019,24(4):562-572. WU S M,LIU L,FU X D,et al. Human detection and multi-task learning for minority clothing recognition[J].Journal of Image and Graphics,2019,24(4):562-572.
11 程远菲.少数民族服饰的特征提取与识别[D].贵阳:贵州民族大学,2018. doi:10.22606/fsp.2019.34005 CHEN Y F.Feature Extraction and Recognition of Ethnic Costumes[D].Guiyang:Guizhou Minzu University,2018. doi:10.22606/fsp.2019.34005
12 NAWAZ M M T,HASAN R,HASAN M A,et al.Automatic categorization of traditional clothing using convolutional neural network[C]// 2018 IEEE/ACIS 17th International Conference on Computer and Information Science.Singapore:IEEE,2018:98-103. DOI:10.1109/ICIS.2018.8466523
13 SIMONYAN K,ZISSERMAN A.Very Deep Convolutional Networks for Large-scale Image Recognition[Z/OL].[2014-09-04].https://arxiv.org/abs/1409.1556. doi:10.5244/c.28.6
14 SZEGEDY C,VANHOUCKE V,IOFFE S,et al.Rethinking the inception architecture for computer vision[C]// Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas:IEEE,2016:2818-2826. DOI:10.1109/CVPR. 2016.308
15 HUANG G,LIU Z,LAURENS V D M,et al.Densely connected convolutional networks[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition.Honolulu:IEEE,2017:2261-2269. DOI:10.1109/CVPR.2017.243
16 XU K,BA J,KIROS R,et al. Show,Attend and Tell:Neural Image Caption Generation with Visual Attention[Z/OL].[2015-12-10].https://arxiv.org/abs/1502.03044v3
17 WOO S,PARK J,LEE J Y,et al. CBAM:Convolutional block attention module[C]// Proceedings of the European Conference on Computer Vision.Munich:ECCV,2018:3-19. DOI:10.1007/978-3-030-01234-2_1
18 HU J,SHEN L,SUN G.Squeeze-and-excitation networks[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018:7132-7141. doi:10.1109/cvpr.2018.00745
[1] 方于华,叶枫. MFDC-Net:一种融合多尺度特征和注意力机制的乳腺癌病理图像分类算法[J]. 浙江大学学报(理学版), 2023, 50(4): 455-464.
[2] 孙美玲,江山. 多尺度有限元法结合分层网格模拟二维奇异摄动的两端边界层问题[J]. 浙江大学学报(理学版), 2022, 49(5): 564-569.
[3] 祝锦泰,叶继华,郭凤,江蕗,江爱文. FSAGN: 一种自主选择关键帧的表情识别方法[J]. 浙江大学学报(理学版), 2022, 49(2): 141-150.
[4] 王欣, 穆绍硕, 陈华锋. 基于多尺度特征提取网络的图像美学量化评分方法[J]. 浙江大学学报(理学版), 2021, 48(1): 69-73.
[5] 傅颖颖, 张丰, 杜震洪, 刘仁义. 融合图卷积神经网络和注意力机制的PM2.5小时浓度多步预测[J]. 浙江大学学报(理学版), 2021, 48(1): 74-83.
[6] 王协, 章孝灿, 苏程. 基于多尺度学习与深度卷积神经网络的遥感图像土地利用分类[J]. 浙江大学学报(理学版), 2020, 47(6): 715-723.
[7] 曲比伟石, 赵振斌, 邓元杰, 张熠. 成都市主城区“两抢一盗”犯罪的多尺度时空格局研究[J]. 浙江大学学报(理学版), 2019, 46(6): 745-754.
[8] 郑嘉丽, 张丰, 杜震洪, 来丽芳, 刘仁义, 刘尧. 传染病的多尺度时空特征分析——以杭州市淋病、细菌性痢疾和流行性腮腺炎为例[J]. 浙江大学学报(理学版), 2018, 45(5): 605-616.