Speaker: David W. Jacobs, Professor in the Department of Computer Science and UMIACS, University of Maryland
Talk Title: 3D Reconstruction: Leveraging Synthetic Data for Lightweight Reconstruction
Abstract: Reconstruction of 3D structure is an important problem with many applications, but in such tasks it is challenging to obtain large amounts of accurately labeled real training data. In many cases, computer graphics provides access to large quantities of labeled data, but there is a domain gap between real images and images generated by graphics. I’ll discuss a series of works that address the challenge of using labeled synthetic data to infer properties of the world from real images.
Bio: David W. Jacobs is a professor in the Department of Computer Science at the University of Maryland with a joint appointment in the University's Institute for Advanced Computer Studies (UMIACS). He received a PhD from MIT and then conducted research at the NEC Research Institute, until he joined the CS department at the University of Maryland. He is currently on leave as an AI Scientist at Meta.
