Research On Deep Learning Based Measurement Method Of Human Body Parameters

Human body parameters measurement has a wide range of applications in clothing customization,ergonomics and many other fields.Traditional methods of measuring human body parameters have many shortcomings,including the long time required and the accuracy depends on experience of people,which cannot achieve arbitrarily,accurate and large-scale repeated measurement.In recent years,research on non-contact human body parameters measurement methods has gradually increased,and many innovative measurement methods are proposed.In this paper,the existing non-contact human body parameters measurement methods are studied,human body contours extraction,human body feature points extraction and human body parameter measurement model construction are optimized,a human body parameters measurement method based on deep learning is proposed.The feasibility and accuracy of the method are proved by experimental verification,and the method is applied in the actual scene.The main contents of this paper are as follows:(1)Body contours extraction methods are studied,and the advantages and disadvantages of all methods are analyzed.Combined with neural network knowledge,an innovative deep learning-based human body contours extraction method is proposed.(2)Feature points extraction methods are studied,and the advantages and disadvantages of all methods are analyzed.Combine prior knowledge and contours characteristics,a human feature points extraction method is proposed.(3)With the contours extraction method and feature points extraction,combined with human knowledge,the human body is divided into six categories,and correlation analysis is carried out to select appropriate model independent variables and adopt new functions.Regression analysis,and finally constructed human body parameters measurement model for each category;(4)311 sample data were collected,and experiments were designed to verify the measurement accuracy of the human body parameters measurement method based on deep learning.In addition,the method is applied in a clothing software.After experimental verification and application in the actual scene,the human body parameters measurement method based on deep learning has proved with high measurement accuracy,and can achieve the goals of arbitrary measurement,accurate measurement,and large-scale repeated and multi-part measurement.