Jiajun Wu

• Stanford AI Lab (SAIL)
• Stanford Vision and Learning Lab (SVL)
• Stanford Computer Graphics Lab
• Stanford Robotics Center (SRC)
• Human-Centered Artificial Intelligence (HAI)

I am an Assistant Professor of Computer Science and, by courtesy, of Psychology at Stanford University. Before joining Stanford, I was a Visiting Faculty Researcher at Google Research, working with Noah Snavely. I finished my PhD at MIT, advised by Bill Freeman and Josh Tenenbaum, and my undergraduate degrees at Tsinghua University, working with Zhuowen Tu.

Research

My group studies physical scene understanding---building machines that see, reason about, and interact with the physical world. Besides learning algorithms, what are the levels of abstraction needed by AI systems in their representations, and where do they come from? Our research aims to answer these fundamental questions, drawing inspiration from nature, i.e., the physical world itself, and from human cognition. Representative projects include Galileo, MarrNet, the Neuro-Symbolic Concept Learner, Neural Flow Maps, and 3D-Fauna.

• multi-modal perception from visual, acoustic, and tactile signals, as in ObjectFolder and RealImpact;
• visual generation of the 4D, physical world, as in 3D-GAN, pi-GAN, Point-Voxel Diffusion, SDEdit, and WonderJourney;
• visual reasoning via physical concept grounding, often in a neuro-symbolic way, as in NS-VQA, Shape Programs, CLEVRER, and LEFT;
• robotics and embodied AI using the learned physical scene representations, as in RoboCook and BEHAVIOR.