TryItOut : Machine Learning Based Virtual Fashion Assistant

Abstract

Image-based virtual try-on systems for fitting new in-shop clothes into a person image have attracted increasing research attention yet is still challenging. They are a future shopping method which can transform the way users shop. They not only change the target clothes into the most fitting shape seamlessly but also preserve the clothes identity such as texture, embroidery, prints etc. in the generated image. In this study, Generative adversarial networks (GAN) Model has been explored for generation of the clothing image and try-on image using CVPR Dataset. A novel approach to generate different poses using the state-of-the art Look into Person (LIP) Parser Model and superimposing the target cloth image. The segmentation of different clothing types was also done in order to identify the texture and clothing type of the person’s clothes and performs well for the images containing obstructions too. The proposed model overcome the limitations of low quality and clear background input images.