基于深度LSTM神经网络的人体服装压力信息预测  

Prediction of Human Clothing Pressure Information Based on Deep LSTM Neural Network

作　　者：韩韬 郝矿荣[1] 丁永生[1] 唐雪嵩 HAN Tao;HAO I(uangrong,DING Yongsheng,TANG Xuesong（College of Information Science and Technology,Donghua University,Shanghai 201620,China)

机构地区：[1]东华大学信息科学与技术学院,上海201620

出　　处：《东华大学学报（自然科学版）》2018年第5期755-761,共7页Journal of Donghua University(Natural Science)

基　　金：科技部国家重点研发计划资助项目(2016YFB0302701);国家自然科学基金资助项目(61473077;61473078);上海市科学技术委员会国际合作资助项目(16510711100);中央高校基本科研业务费专项资金资助项目(2232015D3-32);教育部长江学者奖励计划资助项目(2015-2019)

摘　　要：提出了一种深度长短时记忆(LSTM)神经网络结构,用以在穿着运动紧身衣时,由人体某一个部位的压力信息预测其他几个关键部位的压力信息,进而实现人体上半身全部关键点的压力信息数据预测。通过采用柔性压力阵列收集不同人体穿着不同材质和尺寸的运动紧身衣时的压力信息,在配置图形处理器(GPU)的工作站上得出较好的运算结果。实验结果证明,所提出的深度LSTM神经网络结构比其他算法具有更高的预测准确率。A new deep long short-term memory （LSTM） neural network structure was proposed to predict the pressure information of the other critical parts of the human body based on the information of one part when wearing sport tight clothes. And then, the pressure information of all key points of the upper part of the human body could be obtained by acquiring the pressure information of one part with the human body. By using the flexible pressure array to collect the pressure information of different body wear sports tights of different materials and sizes, a better results are obtained on a workstation equipped with a graphics processing unit （GPU）. Experimental results show that the proposed deep LSTM neural network structure has higher prediction accuracy than other algorithms.

关 键 词：长短时记忆 深度学习 循环神经网络 人体模型 服装压力 

分 类 号：TS941[轻工技术与工程—服装设计与工程]