Robotic Stylist: A design oriented apparel recommendation system

This paper aims to prototype a design-oriented apparel recommendation system based on Artificial Intelligence technology. The Robotic Stylist recommends appropriate clothes to match up with the wearer’s body images and occasions according to the design features of apparel in terms of lines, colors, patterns, prints and textures. Such a system deals with webbased recommendation with real-time results from huge online apparel market to improve users’ experiences while shopping online. A large design evaluation dataset is collected from both fashion experts and peer groups of users via crowdsourcing platform. Artificial Neural Networks are adopted to simulate product judgments process of human brain by training the dataset. The optimization of predicted evaluation results from networks is the solution of recommendations.