Neural network forecasting evaluation of virtual coats' profile appearance

Abstract

In this paper, experts have subjectively evaluated profile appearance of 36 garments which were made of different materials and had different ease. The coherence of experts' evaluations were tested through Friedman method. Main factors witch affected garments appearance ease were used as independent variables to establish BP neural network forecast model and multi-regression forecast model of evaluation toward garments profile appearance. It was found that neural network model was more precise than the multi-regression model.