Online Garment Custom Human Body Parameter Acquisition Model Based On Depth Image

With the development of the Times and the rapid prograss of science and technology,people's material living standard has improved,it will not simply be satisfied with food and clothing,but towords a better material level and higher spiritual needs for development.Clothing as the top needs is indispensable,and how to dress is even one of the daily tasks that people must do.With the Internet as the medium,modern society has transmitted a great deal of information to people,and social networks represented by e-commerce systems are constantly undergoing new changes.All kinds of explosive clothing in the past have been unable to meet people's needs now,even small-volume fashion clothing can not fit people's requirements for personalized clothing.At present,customization of clothing needs to know the exact physical parameters of consumers,and the mainstream clothing dealers of e-commerce platform can only provide standard clothing sizes and cannot help consumers measure.Even though some online marketers solve the problem by letting users provide the size of the initiative,the process of collecting the size of the user itself is often due to non-standard,resulting in unsuitable clothing.In order to solve the problem of inaccurate acquisition of human parameters,more and more researchers have begun to conduct related research work,it is a relatively common solution to allow users to extract the user's human body parameters by taking color photographs after wearing special clothing.However,this method has more special requirements for the user's special wear and environment,and has many restrictions.Therefore,in this study,we use the depth camera Kinect as a camera which an ordinary merchant can afford to get the depth image data.All depth image we get are attached with every user's own body characteristics.Then this study builds an integrated learning model based on convolutional neural network to train the depth image acquired.In order to verify whether the model is more accurate in predicting human parameters,a traditional single neural network is used as the experimental control group.The final experimental results show that the model established in this study is smaller than the traditional single neural network in the prediction of the human parameters,and the maximum error between the predicted value and the actual value obtained is controlled at 0.4cm.Inside,the average error is basically 0.25 cm,which proves that the model has a good improvement in its performance,has better accuracy in the prediction of human parameters,and has a higher practical value.