Welcome to the Visual Intelligence Lab
Welcome to the Visual Intelligence Lab! We study both human and machine vision, with a focus on leveraging the strengths of human vision to inspire the next generation of artificial intelligence (AI) — a perspective we call “strong cognition.” We believe that human vision is remarkable, allowing humans to answer questions that are more insightful than the classic problem of detecting what and where. Many of the sophisticated mental concepts that make up human intelligence have their origins in vision, including intuitive theories of causality, physics, and agency. These theories grounded in vision are essential building blocks of human “commonsense.” Their commonsense nature can be understood from three perspectives: (a) they are commonly shared by humans (and many other animals), (b) knowledge afforded by a visual scene is commonly shared by all observers viewing that scene, and (c) rich visual scene understanding is not constrained to a particular task, but can be commonly applied to solve many tasks. As a result, visually-grounded commonsense forms the foundation for a wide range of intelligent human behavior, including planning, reasoning, problem solving, language, and even morality. Our lab’s goal is to build AI with human-like commonsense knowledge which — just by sharing the same visual environment — can understand, cooperate, and communicate with humans in intuitive, effective, and trustworthy ways.
