PI: Prof. Dr. Michael Bernstein, Camille Utterback

Abstract

As the design thinking process for an idea matures from “getting the right design” to “getting the design right”, attention turns to details of fit, design patterns, and brand. However, large design teams struggle to apply consistent design patterns and style across the entirety of a product, resulting in inconsistent use of design patterns and visual design. In response, we propose Stylo Guide, a system to support distributed design teams in maintaining a consistent design language. Stylo Guide applies deep learning classification techniques to identify when a design deviates from the style guide being used by others on the team. If the design has deviated from the style guide, Stylo Guide highlights areas of deviation and calls out examples from the style guide to help the designer understand the situation and decide whether to revise. We will evaluate Stylo Guide through an experiment where remote design teams are given randomized access to Stylo Guide or to a traditional style guide, and we measure the consistency of the resulting design.