Research On The Relationship Between Posture And Folds Of Virtual Clothing Network Display Based On Machine Learning

In recent years,with the rapid development of computer networks and the accelerating pace of people’s lives,online shopping has become the main way of people’s consumption.As a carrier of human civilization,clothing has become an important part of online shopping products.The display of clothing in the network platform is the main channel for consumers to obtain clothing information.Clothing details directly affects the consumer’s purchase desire.Compare with the traditional clothing display modes,3D virtual clothing can fully displays the details of the garment and giving consumers a new shopping experience.The key of 3D virtual clothing display is the performance of fabric folds.The widely used physical simulation method is realistic,but the process is complex and affects the display efficiency of clothing.Through the data-driven technology to find the relationship between the posture and the clothing folds,and establish a relationship model to achieve the rapid presentation of clothing folds.Take real shirt and sweatpants as research subjects and get joints motion videos.Get 100single-frame images from eight joint point motions that affect garment folds.Extracting joint angle data and fold data from pictures as a sample data set.Eight machine learning algorithms were selected,and the algorithm programs were written with Python language.The relationship between the posture data and the clothing folds data of the eight joint points is learned to obtain the relationship models.The comparison found that the test value and the actual value tend to be consistent and the mean square error is controlled within an acceptable range,so those models were determined to be the final models.