Advances in deep learning enabled sampling realistic images via generative modeling. This leads to new avenues in visual design and content creation, e.g. in fashion, where visualization is a key component. GANs can be used to create personalized visual content - e.g. rendering an outfit on a human body and creating unique designs - which can enrich shopping experience on e-commerce platforms. We will demo two projects, where we used GANs to create fashion images and enable novel applications:
We work on generating high-resolution images of fashion models wearing desired outfits and standing in different poses. At Zalando, we provide quality photographs of fashion models wearing the articles in our online selection. These photographs help customers visualise the garments they browse and enhance the shopping experience. But what if our customers wish to visualise an individually created outfit? Zalando has a large and evolving assortment of garments, which makes it infeasible to photograph all outfit combinations. To solve this challenge, we work on a “Fashion Renderer”, which creates a computer-generated image of a fashion model wearing an input outfit for an input body pose.
Fashion customers often have a visual idea of what they would like to buy. However, finding the right article can be a time-consuming process, as people need to convert their visual ideas into accurate linguistic search terms, and search engines should correctly interpret customers’ search queries and retrieve relevant results. We enable search in a visual-only space by allowing customers to generate and breed different dress designs with using a style-based GAN. Created designs are used as a visual query to retrieve existing dresses in real time. This approach attempts to eliminate representation and interpretation problems in the word-based search and provides a novel way for searching fashion items.