Automatic Generation of Matching Clothes Design Using Generative Adversarial Networks

摘要

This paper introduces a new automated system for designing new clothes. Given sample clothes, this system generates matching cloth designs in a few seconds. Optimizing the new cloth design times and decreasing the cost of such designs are some of the main desires of the textile industry, which can be achieved with the help of the system proposed. Our system employs Generative Adversarial Networks of deep learning techniques, which became very popular for many applications. The proposed system has also some valuable properties for the consumers, such as having a completely original design that no other clothing brand offers. We validated the proposed system by performing experiments on volunteers by showing the produced designs and getting their opinions. We are very encouraged by the initial results and we think that the system may be applicable for practical employment.