Design of Clothing Collocation Model Based on Expert Opinion

Abstract

The characteristic similarity of clothing can describe the degree of similarity between clothing and clothing, from this point of view can be excavated more clothing collocation rules. In this paper, the garment collocation algorithm is studied based on deep learning technology. On the basis of the clothing collocation style of expert opinions, the expert collocation rules are expanded through the characteristic similarity of clothing. The Convolution Neural Network (CNN) and the Long Short-Term Memory Network (LSTM) in deep learning can be used to find the characteristics between clothing from the perspective of the style features, compatibility features and semantic features of clothing. A new collocation combination can be found by the similarity calculation of the characteristics. Through the simulation experiment and the judgment of the evaluation index, the prediction performance of the algorithm in this paper is better than Siamese Convolution Neural Network (SCNN) and the association rule algorithm of and data transaction.