基于XGBoost算法对新疆女性臀部体型判别及原型修正

doi:10.13475/j.fzxb.20191002107

摘要/Abstract

摘要：

为准确判断下体体型,使新疆地区女性裙装更加合体,分析了220名18~25岁的新疆地区青年女性臀部数据,通过因子分析与相关指数得到聚类指标:后臀长腰围比、后臀长臀围比,采用K-means聚类方法将臀部分为3类,使用Python软件建立XGBOOST臀部判别模型。首先对不同算法模型进行比较分析表明,运用XGBOOST方法的测试集精准度最高为98.4%。其次修正新疆地区中间体的裙装原型,发现其后臀长与标准裙装原型后臀长相差2.4 cm,说明新疆地区女性臀部相比国内其他地区偏翘。将判别算法运用到数据系统中,可提高体型判别效率,为人体相关领域提供数据支持。

关键词: 新疆青年女性, 臀部体型分类, 体型, XGBOOST算法, 裙装原型修正

Abstract:

This paper addresses the fitting issues for Xinjiang skirts through a study on the size of lower body shape. In order to determine the true shape of hip body accurately, hip data of 220 young Xinjiang females aged 18 to 25 were analyzed. Using the factor analysis and correlation coefficients mergods, 2 factor for clustering were obtained, i.e. back hip length to waist girth ratio, back hip length to hip girth ratio, and 3 hip body types were defined based on the K-means clustering method. On this basis, Python software was used to establish an XGBoost buttock discriminant model. The work involved three parts. The first thing was to compare and analyze different algorithm models and XGBoost was used to evaluate the accuracy, and the highest accuracy reached 98.4%. The second part was to,modify the intermediate skirt prototype for young Xinjiang females, where it was found that the rear hip length was 2.4 cm different from the rear hip length of the standard dress prototype, indicating that the young Xingjiang female hips is larger than that in other regions. Finally, the discriminating algorithm was applied to the data system. The outcome of the research is shown to have improved discrimination efficiency for body size, and the method provides data support for other human related fields.

Key words: young Xinjiang female, hip shape classification, body shape, XGBoost algorithm, skirt prototype correction

中图分类号:

TS941.17

刘婷婷, 徐红, 梅馨元, 刘一心, 肖爱民. 基于XGBoost算法对新疆女性臀部体型判别及原型修正[J]. 纺织学报, 2020, 41(07): 147-153.

LIU Tingting, XU Hong, MEI Xinyuan, LIU Yixin, XIAO Aimin. Young Xinjiang female hip shape characterization and prototype correction using XGBoost algorithm[J]. Journal of Textile Research, 2020, 41(07): 147-153.

图/表 11

图1

表1

表2

表3

表4

图2

表5

图3

表6

图4

图5

参考文献 20

[1]	尹玲, 张文斌, 许才国. 基于有序样本最优分割法的女性体型分类[J]. 纺织学报, 2014,35(9):114-119.
	YIN Lin, ZHANG Wenbin, XU Caiguo. Female body shape classification based on optimal segmentation method for orderly samples[J]. Journal of Textile Research, 2014,35(9):114-119.
[2]	MAJA Mahnic Naglic, SLAVENKA Petrak. A method for body posture classification of three-dimensional body models in the sagittal plane[J]. Textile Research Journal, 2017,89(2):133-149.
[3]	王军, 李晓久, 潘力, 等. 东北地区青年女性腰臀部体型特征与分类[J]. 纺织学报, 2018,39(4):106-110.
	WANG Jun, LI Xiaojiu, PAN Li, et al. Waist hip somatotype and classification of young women in Northeast China[J]. Journal of Textile Research, 2018,39(4):106-110.
[4]	石小强, 王宏付. 青年女性臀部高度特征分析[J]. 纺织学报, 2015,36(5):74-78.
	SHI Xiaoqiang, WANG Hongfu. Analysis on characteristics about hip height of young female[J]. Journal of Textile Research, 2015,36(5):74-78.
[5]	尹玲, 夏蕾, 许才国. 基于随机森林的女性体型判别[J]. 纺织学报, 2014,35(5):113-117.
	YIN Lin, XIA Lei, XU Caiguo. Female body shape prediction based on random forest[J]. Journal of Textile Research, 2014,35(5):113-117.
[6]	景晓宁, 李晓久. 朴素贝叶斯算法在女童体型判别中的应用[J]. 纺织学报, 2017,38(12):124-128.
	JING Xiaoning, LI Xiaojiu. Application of Naive Bayesian method in girl's figure discrimination[J]. Journal of Textiles Research, 2017,38(12):124-128.
[7]	CHEN Tianqi, GUESTRIN C. XGBoost: a scalable tree boosting system[C] //Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM Press, 2016: 785-794.
[8]	LI G Y, LI W, TIAN X L, et al. Short-term electricity load forecasting based on the XGBoost algorithm[J]. Smart Grid, 2017,7(4):274-285.
[9]	程朋朋, 陈道玲. 福建青年女性的人体测量与体型特点分析[J]. 北京服装学院学报(自然科学版), 2017(1):93-100.
	CHENG Pengpeng, CHEN Daoling. Analysis of the human measurement and body characteristics of young women in Fujian[J]. Journal of Beijing Institute of Clothing Technology(Natural Science Edition), 2017(1):93-100.
[10]	张文斌. 服装结构设计[M]. 北京: 中国纺织出版社, 2017: 59-69.
	ZHANG Wenbin. Clothing structure design [M]. Beijing: China Textile & Apparel Press, 2017: 59-69.
[11]	三吉满智子. 服装造型学[M]. 北京: 中国纺织出版社, 2006: 54-180.
	TOMOKO Miyoshi. Costume styling [M]. Beijing: China Textile & Apparel Press, 2006: 54-180.
[12]	黄灿艺. 福建地区青年女性体型划分与尺寸分档[J]. 纺织学报, 2012,33(5):111-115.
	HUANG Canyi. Classification of figure type and size grading of young women of Fujian province[J]. Journal of Textile Research, 2012,33(5):111-115.
[13]	SU J, KE Y, KUANG C, et al. Converting lower-body features from three-dimensional body images into rules for individualized pant patterns[J]. Textile Research Journal, 2018,89(11):2199-2208.
[14]	方方, 王子英. K-Means聚类分析在人体体型分类中的应用[J]. 东华大学学报(自然科学版), 2014,40(5):593-598.
	FANG Fang, WANG Ziying. Application of K-means clustering analysis in the body shape classification[J]. Journal of Donghua University(Natural Science), 2014,40(5):593-598.
[15]	PEI Jie, PARK Huiju, SUSAN P Ashdown. Female breast shape categorization based on analysis of CAESAR 3D body scan data[J]. Textile Research Journal, 2018,89(4):590-611. doi: 10.1177/0040517517753633
[16]	吴奕颖, 肖爱民. 新疆青年女性体型研究[J]. 天津纺织科技, 2015(1):63-64.
	WU Yiying, XIAO Aimin. Study on the size of young women in Xinjiang[J]. Tianjin Textile Technology, 2015(1):63-64.
[17]	方方, 徐跃瑄. 基于三维数据的青年女性腰臀部形态分类分析[J]. 服装学报, 2017,2(4):294-300.
	FANG Fang, XU Yuexuan. Morphological classification of the waist-hip shape of young woman based on three-dimensional data[J]. Journal of Clothing Research, 2017,2(4):294-300.
[18]	王燕, 郭元凯. 改进的XGBoost模型在股票预测中的应用[J]. 计算机工程与应用, 2019,24(9):1-10.
	WANG Yan, GUO Yuankai. Application of improved XGBoost model in stock forecasting[J]. Computer Engineering and Applications, 2019,24(9):1-10.
[19]	BINA Abling, KATHLEEN Magglo. Integrating: draping and drawing[M]. America: Fairchild Books Visuals, 2016: 63-70.
[20]	刘瑞璞. 服装纸样设计原理与应用[M]. 北京: 中国织出版社, 2008: 77-97.
	LIU Ruipu. The principles and applications of clothing paper design [M]. Beijing: China Textile & Apparel Press, 2008: 77-97.

相关文章 15

[1]	王婷, 顾冰菲. 基于二维图像的青年女性颈肩部形态自动识别[J]. 纺织学报, 2020, 41(12): 111-117.
[2]	余佳佳, 李健. 中国东部地区青年女性人体体型分类[J]. 纺织学报, 2020, 41(05): 134-139.
[3]	周捷, 毛倩. 基于改进快速搜索和发现密度峰值算法的男童体型分类及判别[J]. 纺织学报, 2020, 41(05): 146-152.
[4]	肖平, 张昭华, 秦王洁, 范佳宜. 网购服装产品属性对顾客合体满意度的影响[J]. 纺织学报, 2019, 40(06): 91-96.
[5]	李小辉张思严赵楚瞳边菲. 衣身肩斜角与人体体型特征的关系[J]. 纺织学报, 2018, 39(11): 111-115.
[6]	王军李晓久潘力姚彤于佐君. 东北地区青年女性腰臀部体型特征与分类[J]. 纺织学报, 2018, 39(04): 106-110.
[7]	景晓宁李晓久. 朴素贝叶斯算法在女童体型判别中的应用[J]. 纺织学报, 2017, 38(12): 124-128.
[8]	姚怡马静吴欢李秦曼邹奉元. 基于小波系数的青年女性体型分类及原型纸样[J]. 纺织学报, 2017, 38(12): 119-123.
[9]	夏凤勤毋戈谢昊洋钟跃崎. 基于人体纵截面特征曲线的体型分类[J]. 纺织学报, 2017, 38(06): 86-91.
[10]	邓椿山李琴周莉张龙琳. 体型分析在观测服装号型适应性上的应用[J]. 纺织学报, 2017, 38(01): 111-115.
[11]	金娟凤庞程方陈伟杰叶晓露邹奉元. 青年男性肩点横截面曲线及其体型细分[J]. 纺织学报, 2016, 37(08): 100-106.
[12]	王祺明. 基于人体三围截面面积的江浙地区女性体型分类[J]. 纺织学报, 2016, 37(05): 131-136.
[13]	刘咏梅张小雪郭云昕. 国内外服装用人体数据库调查与分析[J]. 纺织学报, 2015, 36(06): 141-147.
[14]	尹玲张文斌许才国. 基于有序样本最优分割法的女性体型分类[J]. 纺织学报, 2014, 35(9): 114-0.
[15]	倪世明金娟凤庞程方邹奉元. 基于纵向轮廓曲线的青年女性体型细分研究[J]. 纺织学报, 2014, 35(8): 87-0.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

成分	旋转载荷平方和
成分	特征根	贡献率/%	累积贡献率/%
1	5.616	33.035	33.035
2	3.195	18.797	51.832
3	2.972	17.480	69.311
4	1.326	7.798	77.110

直接变量	成分
直接变量	1	2	3	4
臀围	0.888	0.125	0.173	0.027
体重	0.880	0.244	0.128	0.055
大腿根围	0.861	-0.036	0.054	0.069
腰围	0.834	0.041	0.001	-0.056
腰厚	0.803	0.027	0.015	-0.219
臀厚	0.800	0.045	-0.042	-0.022
中腰围	0.794	0.217	0.103	0.087
腹厚	0.772	0.128	0.006	-0.135
身高	0.201	0.892	0.218	0.055
臀高	0.124	0.884	0.037	0.032
腰高	0.169	0.877	0.305	0.036
膝盖中点高	0.016	0.782	0.042	0.022
后臀长	0.059	0.123	0.964	0.104
前臀长	0.081	0.167	0.962	-0.071
侧臀长	0.077	0.192	0.936	0.061
臀突上角	-0.017	0.120	-0.093	0.796
腰侧角	-0.074	-0.021	0.165	0.762

主要因子	指标	相关指数	变异系数
围度因子	体重	0.525	0.115 5
	臀围	0.523	0.054 0
	大腿根围	0.451	0.073 2
	腰围	0.452	0.080 5
	腰厚	0.401	0.110 6
	中腰围	0.416	0.074 5
	臀厚	0.384	0.085 0
	腹厚	0.380	0.092 3
高度因子	身高	0.598	0.033 8
	臀高	0.513	0.050 5
	腰高	0.594	0.046 1
	膝盖中点高	0.351	0.049 4
臀长因子	后臀长	0.872	0.104 9
	前臀长	0.843	0.098 6
	侧臀长	0.841	0.102 7
角度因子	臀突上角	0.256	0.289 5
角度因子	腰侧角	0.256	0.237 4

项目	指标	BMI值	臀腰差/cm	臀腰比	腰围身高比	臀围身高比	后臀长腰围比	后臀长臀围比	后臀长身高比
F值	体重	2.335	2.311	3.262	5.683	2.907	2.393	3.664	1.752
	腰围	2.847	4.336	18.120	6.558	11.736	5.375	2.328	1.339
	臀围	1.997	2.381	7.826	21.714	1.052	1.689	19.136	1.341
	身高	7.826	1.395	1.656	1.474	1.964	1.766	1.947	10.728
	后臀长	0.950	0.959	0.803	0.634	0.612	12.716	95.599	139.101
P值	体重	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.034
	腰围	0.008	0.000	0.000	0.000	0.000	0.000	0.000	0.170
	臀围	0.001	0.000	0.000	0.000	0.404	0.009	0.000	0.169
	身高	0.000	0.078	0.018	0.113	0.008	0.005	0.002	0.000
	后臀长	0.589	0.546	0.847	0.853	0.917	0.000	0.000	0.000

体型类别	后臀长腰围比	后臀长臀围比	占比/%
1	0.33	0.25	21
2	0.28	0.22	42
3	0.24	0.19	37