Bernal, Guillermo, Lily Zhou, Haripriya Mehta, and Pattie Maes. NeurIPS Workshop on Machine Learning for Creativity and Design, December 14, 2019.

Abstract

Paper Dreams focuses on how current state-of-the-art machine learning algorithm sand multi-modal input technologies can be integrated to augment human creativity and empower a wide range of users. We present a web application that promotes a highly dynamic, interactive back-and-forth between a human artist and thePaper Dreams system, incorporating customized machine learning models and representations to assist the user’s visual expression. The Paper Dreams system processes the sketch the user generates, “recognizes” what has been drawn, and creates personalized suggestions for new elements and colors. Papers Dreams aims to inspire and empower individuals who believe they are not “the creative type”, are experiencing an artistic block, or are looking to express themselves in new ways.