This will be a hybrid event with in-person attendance in Levine 512 and virtual attendance on Zoom.

ABSTRACT

Our group’s goal is to build a world simulator from visual observations that can answer “what-if” questions. In pursuit of this goal, we develop various methods for modeling the world from images and harnessing physical simulation and generative models to create immersive new experiences. In today’s talk, we will focus on creating realistic and immersive experiences that allow users and agents to interact with images, videos, and 3D worlds captured from the real world. We will cover applications in content creation, autonomy, and agriculture. Finally, I will provide a brief personal outlook on open research topics that make all images and videos interactive.