Learning to See and Generate Humans
We teach computers to perceive and understand people and their behavior from visual data. We develop algorithms to synthesize realistic humans in 3D scenes.
We are a research group within the Insitute of Visual Computing of the Department of Computer Science at ETH Zürich. Our research interests lie in computer vision and the combination with machine learning. We work on discovering and proposing algorithms and implementations for solving high-level visual recognition problems. The goal is to advance the frontier of robust machine perception in real-world settings. Our current research interests include:
Human-scene Interaction. Humans constantly interact with the 3D environment around them. To better understand and model the human-scene interactions, we develop fundamental representations, propose novel algorithms and estabilish new benchmarks.
Action and Behavior. We are interesting in understanding the fine-grained details of human action and behavior. We also aim to understand the dynamics, intention and causality of human behaviour from visual data. In particular, we study human action and behavior in the context of real-world 3D environments.
Human Pose Estimation and Tracking. We study optimization problems for the task of people pose estimation and tracking in real-world crowded scenes. We also study novel learning techniques that connect combinatorial optimization with deep neural networks and blur the boundary between these two model classes.