Cognitive characteristics based autonomous development of clothing style

Abstract

Due to the subjective characteristics and rapid change of fashion style, it is relatively hard to predefine the style feature in style classification systems. In this paper we present a cognitive characteristics based clothing style autonomous development model. By the addition of special domain related information to the classic itti visual attention model, we achieve the multi-object attention model of the clothing style. And based on this we implemented the autonomous development of clothing style recognition by Multi-Layer In-place Learning Network (MILN in short). Experiments prove the feasibility and effectiveness of our model.