Body Shape Recognition and Prototype Construction Based on LightGBM Algorithm

Abstract

The regional characteristics of middle-aged women in Xinjiang are obvious, and the phenomenon of incompatible clothing is prominent. In order to avoid the information loss of principal component analysis or factor analysis during body type classification, this study uses the LightGBM algorithm to establish a body type recognition model based on the results of K-Means clustering, and compares it with random forest and linear regression recognition models. It is found that the LightGBM model has the best body type recognition effect. it is good. Through correlation analysis and regression analysis, regression equations between variables are obtained. Based on the new cultural prototype, Liu Ruipu women’s clothing prototype, and Donghua prototype, a clothing prototype suitable for middle-aged women in Xinjiang is established. Through virtual fitting evaluation and actual fitting, the experimental prototypes established based on the Donghua prototype and the Liu Ruipu prototype can meet the needs of middle-aged women for the comfort and aesthetics of clothing. The experimental prototype established by the Donghua prototype has the best effect.