Abstract: The evaluation of the shape of clothing is usually carried out by subjective methods,this methods are usually direct and quick,but lack of quantities analysis.To solve this problem,the shape parameters of semi-fitting skirt are defined as skirt width,thickness,hem pleat node number,average hem radium,Std.deviation of average hem radium and hem area.The linear regression model and artificial neural networks model for the prediction of semi-fitting skirt shape are built up separately through fabric properties.By analyzing the relative coefficient between the measurement value and predictive value of these modes,it is proved that the precision of network model is better than that of linear regression model.