Graphical user interface for image inpainting

Abstract

Image inpainting is a type of image conservation process where missing or removed portions of an image are filled in by artificially generating a plausible result. This project is based on the existing Pluralistic Image Completion repository which implements an image inpainting model that can generate a diverse range of different results. The current repository presents its interface as a PyQt5 python application that is run on the local desktop. The goal of this project is to design and implement a web-based graphical user interface (GUI) in order to improve on the visual aspect, usability and functionality of the existing GUI. In order to create a new GUI, user evaluations were first conducted to find out how the user experience can be improved and what users are looking for in a new interface. Usability measurements will be recorded in order to find out a quantitative benchmark from which comparisons can be made. Ben Shneiderman’s principles of good interface design will also be used in designing the new GUI, in order to reduce user frustrations and limitations in the current interface. Finally, a second round of user studies will be conducted to evaluate whether the new interface has resolved user issues and allowed for a better user experience. Further possible extensions to the project will also be explored, while limitations will be identified to consider all possible biases in the research process.