多尺度SE-Xception服装图像分类

doi:10.3785/j.issn.1008-973X.2020.09.008

浙江大学学报(工学版)

2020, Vol. 54

Issue (9): 1727-1735 DOI: 10.3785/j.issn.1008-973X.2020.09.008

计算机技术

多尺度SE-Xception服装图像分类

陈巧红(

),陈翊,李文书,贾宇波

浙江理工大学信息学院，浙江杭州 310018

Clothing image classification based on multi-scale SE-Xception

Qiao-hong CHEN(

),YI CHEN,Wen-shu Li,Yu-bo JIA

School of Information Science and Technology, Zhejiang Sci-Tech University, Hangzhou 310018, China

全文: PDF(971 KB) HTML

摘要：

应用当前较新颖且分类性能靠前的卷积神经网络Xception作为基础网络结构，尝试采用多尺度的深度可分离卷积来提升模型特征信息的丰富度，在模型中嵌入SE-Net模块增强有用特征通道，减弱无用特征通道. 实验结果表明：提出的多尺度SE-Xception模型在2种噪声程度不同的服装数据集中均取得不错表现；ACS数据集的平均分类准确率为78.34%，分别高于VGG-16、ResNet-50和Xception模型8.52%、4.81%、3.69%；验证了多尺度SE-Xception模型具有更好的特征提取能力，能够提取到更多的服装信息，从而提高服装图像分类效果，一定程度上解决了特征尺度单一、信息丰富度低的问题.

关键词： 服装图像分类; 多尺度SE-Xception; 图像识别; 深度学习; 机器学习; 卷积神经网络（CNN）

Abstract:

The current newer and better convolutional neural network Xception was used as the foundation network structures. Multi-scale depth separable convolution was employed to improve the richness of model feature information. The SE-Net model was embedded in the model to enhance the useful feature channels and weaken the useless feature channels. The experimental results show that the multi-scale SE-Xception model achieved good performance in two different noise clothing datasets. The average classification accuracy of ACS dataset was 78.34%, which was higher than VGG-16, ResNet-50 and Xception models by 8.52%, 4.81% and 3.69%, respectively. Therefore, it’s verified that the multi-scale SE-Xception model has better ability to extract features, can obtain more clothing information, and thus improve the clothing image classification effect. To some extent, a multi-scale SE-Xception network is conducive to solve the single feature scale and low information richness problems.

Key words: clothing image classification multi-scale SE-Xception image identification deep learning machine learning convolutional neural network (CNN)

收稿日期: 2019-08-14 出版日期: 2020-09-22

CLC:

TP 181

基金资助: 国家自然科学基金资助项目（51775513）

作者简介: 陈巧红（1978—），女，副教授，从事计算机辅助设计及机器学习技术研究. orcid.org/0000-0003-0595-341X. E-mail： chen_lisa@zstu.edu.cn

	服务
	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	作者相关文章
	陈巧红
	陈翊
	李文书
	贾宇波

引用本文:

陈巧红,陈翊,李文书,贾宇波. 多尺度SE-Xception服装图像分类[J]. 浙江大学学报(工学版), 2020, 54(9): 1727-1735.

Qiao-hong CHEN,YI CHEN,Wen-shu Li,Yu-bo JIA. Clothing image classification based on multi-scale SE-Xception. Journal of ZheJiang University (Engineering Science), 2020, 54(9): 1727-1735.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2020.09.008 或 http://www.zjujournals.com/eng/CN/Y2020/V54/I9/1727

图 1 Xception模型的主要网络模块示意图

图 2 深度可分离卷积的结构图

图 3 SE-Net模块流程图

图 4 多尺度深度可分离卷积的结构图

表 1 多尺度SE-Xception模型的整体框架

图 5 ACS数据集与DeepFashion数据集示例图

图 6 不同模型在ACS数据集上的损失变化曲线对比

图 7 不同模型在ACS数据集上的准确率变化曲线对比

表 2 不同组合的卷积核实验结果对比

表 3 Xception模型与多尺度SE-Xception模型在不同数据集上的实验结果对比

表 4 不同模型在ACS数据集上的平均分类准确率对比

1	中国电子商务研究中心. 2015—2016年度中国服装电商行业报告[R/OL]. (2016-12-26)[2018-12-17]. http://www.100ec.cn/zt/fzyxbg/.
2	OJALA T, PIETIKAINEN M, HARWOOD D. Performance evaluation of texture measures with classification based on Kullback discrimination of distributions [C] // Proceedings of 12th International Conference on Pattern Recognition. Jerusalem: IEEE, 1994: 582-585.
3	LOWE D G Distinctive image features from scale-invari-ant keypoints[J]. International Journal of Computer Vision, 2004, 60 (2): 91- 110 doi: 10.1023/B:VISI.0000029664.99615.94
4	DALAL N, TRIGGS B. Histograms of oriented gradients for human detection [C] // Computer Vision and Pattern Re-cognition. San Diego: IEEE, 2005: 886-893.
5	BOSSARD L, DANTONE M, LEISTNER C, et al. Apparel classification with style [C] // Asian Conference on Com-puter Vision. Daejeon: Springer, 2012: 321-335.
6	HUO P, WANG Y, LIU Q. A part-based and feature fus-ion method for clothing classification [C] // Pacific Rim Conference on Multimedia. Xi’an: Springer, 2016: 231-241.
7	SURAKARIN W, CHONGSTITVATANA P. Predicting types of clothing using SURF and LDP based on bag of features [C] // International Conference on Electrical Engineering/Electronics. Hua Hin: IEEE, 2015: 1-5.
8	吴苗苗, 刘骊, 付晓东, 等款式特征描述符的服装图像细粒度分类方法[J]. 计算机辅助设计与图形学学报, 2019, 31 (5): 780- 791 WU Miao-miao, LIU Li, FU Xiao-dong, et al Fine-grained clothing image classification by style feature description[J]. Journal of Computer-aided Design and Computer Graphics, 2019, 31 (5): 780- 791 doi: 10.3724/SP.J.1089.2019.17380
9	LIU Z, LUO P, QIU S, et al. DeepFashion: Powering ro-bust clothes recognition and retrieval with rich annotations [C] // 2016 the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 1096-1104.
10	DONG C Y, SHI Y Q, TAO R. Convolutional neural ne-tworks for clothing image style recognition [C] // 2018 International Conference on Computational, Modeling, Simulation and Mathematical Statistics. Pennsylvania: CMSMS, 2018: 592-597.
11	包青平, 孙志锋基于度量学习的服装图像分类和检索[J]. 计算机应用与软件, 2017, 34 (4): 255- 259 BAO Qing-ping, SUN Zhi-feng 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
12	张振焕, 周彩兰, 梁媛基于残差的优化卷积神经网络服装分类算法[J]. 计算机工程与科学, 2018, 40 (2): 354- 360 ZHANG Zhen-huan, ZHOU Cai-lan, LIANG Yuan An optimized clothing classification algorithm based on residual convolutional neural network[J]. Computer Engineering and Science, 2018, 40 (2): 354- 360 doi: 10.3969/j.issn.1007-130X.2018.02.023
13	厉智, 孙玉宝, 王枫, 等基于深度卷积神经网络的服装图像分类检索算法[J]. 计算机工程, 2016, 42 (11): 309- 315 LI Zhi, SUN Yu-bao, WANG Feng, et al Clothing image classification and retrieval algorithm based on deep convolutional neural network[J]. Computer Engineering, 2016, 42 (11): 309- 315 doi: 10.3969/j.issn.1000-3428.2016.11.053
14	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks [C] // Advances in Neural Information Processing Systems 25 (NIPS 2012). Nevada: NIPS, 2012: 1097-1105.
15	SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition [EB/OL]. (2015-04-10)[2018-12-17]. https://arxiv.org/abs/1409.1556.
16	SZEGEDY C, LIU W, JIA Y, et al. Going deeper with convolutions [C] // 2015 the IEEE Conference on Computer Vision and Pattern Recognition. Boston: IEEE, 2015: 1-9.
17	HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition [C] // 2016 the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 770-778.
18	CHOLLET F. Xception: deep learning with depthwise separable convolutions [C] // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 1251-1258.

[1]	许佳辉,王敬昌,陈岭,吴勇. 基于图神经网络的地表水水质预测模型[J]. 浙江大学学报(工学版), 2021, 55(4): 601-607.
[2]	王虹力,郭斌,刘思聪,刘佳琪,仵允港,於志文. 边端融合的终端情境自适应深度感知模型[J]. 浙江大学学报(工学版), 2021, 55(4): 626-638.
[3]	张腾,蒋鑫龙,陈益强,陈前,米涛免,陈彪. 基于腕部姿态的帕金森病用药后开-关期检测[J]. 浙江大学学报(工学版), 2021, 55(4): 639-647.
[4]	战友,李强,马啸天,王郴平,邱延峻. 基于宏微观纹理特征融合的路面摩擦性能预测[J]. 浙江大学学报(工学版), 2021, 55(4): 684-694.
[5]	郑英杰,吴松荣,韦若禹,涂振威,廖进,刘东. 基于目标图像FCM算法的地铁定位点匹配及误报排除方法[J]. 浙江大学学报(工学版), 2021, 55(3): 586-593.
[6]	于勇,薛静远,戴晟,鲍强伟,赵罡. 机加零件质量预测与工艺参数优化方法[J]. 浙江大学学报(工学版), 2021, 55(3): 441-447.
[7]	徐利锋,黄海帆,丁维龙,范玉雷. 基于改进DenseNet的水果小目标检测[J]. 浙江大学学报(工学版), 2021, 55(2): 377-385.
[8]	许豪灿,李基拓,陆国栋. 由LeNet-5从单张着装图像重建三维人体[J]. 浙江大学学报(工学版), 2021, 55(1): 153-161.
[9]	黄毅鹏,胡冀苏,钱旭升,周志勇,赵文露,马麒,沈钧康,戴亚康. SE-Mask-RCNN：多参数MRI前列腺癌分割方法[J]. 浙江大学学报(工学版), 2021, 55(1): 203-212.
[10]	郑浦,白宏阳,李伟,郭宏伟. 复杂背景下的小目标检测算法[J]. 浙江大学学报(工学版), 2020, 54(9): 1777-1784.
[11]	周登文,田金月,马路遥,孙秀秀. 基于多级特征并联的轻量级图像语义分割[J]. 浙江大学学报(工学版), 2020, 54(8): 1516-1524.
[12]	张彦楠,黄小红,马严,丛群. 基于深度学习的录音文本分类方法[J]. 浙江大学学报(工学版), 2020, 54(7): 1264-1271.
[13]	明涛,王丹,郭继昌,李锵. 基于多尺度通道重校准的乳腺癌病理图像分类[J]. 浙江大学学报(工学版), 2020, 54(7): 1289-1297.
[14]	闫旭,范晓亮,郑传潘,臧彧,王程,程明,陈龙彪. 基于图卷积神经网络的城市交通态势预测算法[J]. 浙江大学学报(工学版), 2020, 54(6): 1147-1155.
[15]	杨萍,王丹,康子健,李童,付利华,余悦任. 基于模式识别和集成CNN-LSTM的阵发性房颤预测模型[J]. 浙江大学学报(工学版), 2020, 54(5): 1039-1048.

Viewed

Full text

Abstract

Cited

Shared

Discussed