Computer vision based personalized clothing assistance system: A proposed model

With fashion industry included e-commerce (worldwide) is expected to hit the $35 billion mark by 2020. There's a need for applications which help the user in making intelligent decisions on their day to day online purchases. In this paper, our aim is to build a system that would be able to understand fashion and the user to provide personalized clothing recommendations to the user. Our approach includes Caffe, a deep learning framework for computer vision tasks such as Clothing type classification and Clothing attribute classification. Furthermore we use Conditional Random Fields (CRF) to learn the intricacies of fashion. CRFs also learn the correlations between attributes of the user such as ethnicity, body type etc., expert opinion and the type of outfit. We expect the proposed system would be able to provide personalized recommendations.