Facial Composite Generation from Natural Language Text Descriptions using Stacked Generative Adversarial Networks

Abstract

Sketch artistry has been a part of law enforcement since the late 1800s with documented use in wanted posters from the American West and Great Britain. More recently, computerized forms of facial composite generators have taken a presence in law enforcement by building faces from individually described features. I propose a new stacked generative adversarial network model for generating facial composites. This solution focuses on minimizing difficulty for an operator to utilize the model, and maximizing the potential for the user to engage in the psychological process of recognition rather than recall.

Michael J. Ryan

Masters student in NLP

My research interests include multilingual NLP and memory/knowledge modeling