Chris Sweeney, a Ph.D. student in the Four Eyes Lab at the University of California, Santa Barbara, will speak on “Continuous, Dynamic Reconstruction of the 3D World” on Friday, September 26, at 3:30pm in Rice Hall 130.
He will discuss how structure from motion techniques allows reconstruction of a 3D model of the world from public photos from a variety of sources, including Flickr, Panoramio, and Facebook. The resulting 3D model is useful for applications such as urban planning, emergency response, virtual tourism, and augmented reality. However, current models assume a static environment despite the fact that the real world is constantly changing and evolving. Updating or extending large-scale 3D models currently requires rebuilding them, which is computationally costly and not feasible when there are constant streams of updates. As a result, applications that require real-time access to the 3D models such as outdoor augmented cannot be used reliably with current structure from motion methods because the visual information may be stale.
He will also review the state of the art of large-scale structure from motion and discuss his current work towards creating efficient, dynamic models that can be crowd-sourced continuously at scale. He will present a method for calibrating unknown camera internal parameters at scale, followed by a unified formulation for expressing many image observations as one, and explain our efficiently merging large reconstructions and our ongoing and future work.
Sweeney is visiting UVA as part of the Department of Computer Science's Alumni Colloquium. He received his B.S. in Computer Science from UVA in 2011. He has received an NSF Graduate Research Fellowship, a UCSB Graduate Opportunity Fellowship, a Google Outstanding Research Scholarship and a Best Paper award at the International Symposium for Mixed and Augmented Reality (ISMAR) 2012. He is currently a visiting PhD student at ETH-Zurich working in the Computer Vision and Geometry Lab under Prof. Marc Pollefeys.