- Reconstruct and Visualize the World
平成27年7月30日 (木) 13時30分〜15時 東京大学理学部七号館007講義室
Automated 3D reconstruction and visualization techniques from images, a field known as 3D photography or 3D computer vision, have gone through a revolution in the last decade. With state-of-the-art techniques, one can now download millions of images from Internet by typing a keyword, identify clusters of images observing the same scenes, and reconstruct detailed 3D models fully automatically. The advent of high quality consumer-grade depth sensors, such as Microsoft Kinect Camera, is also an exciting new element in the field with much potential. In addition to being an active research topic in Computer Vision and Computer Graphics, 3D photography techniques have been extensively deployed as real products in industry, and have been used in a variety of other fields such as archaeology and civil engineering. These techniques are also becoming a driving factor for up-coming technological innovations in 3D printing, augmented reality, virtual reality, autonomous driving, drones, and robotics. In this talk, I will present a variety of 3D reconstruction and visualization techniques, ranging from small-scale object reconstruction in a lab environment, city-scale outdoor scene reconstruction from millions of images, to indoor scene reconstruction of entire buildings from laser range sensors.
Yasutaka Furukawa is an assistant professor at Washington University in St. Louis. His research is in automated 3D reconstruction and visualization techniques from images. His multi-view stereo (MVS) algorithm has been recognized as the best 3D reconstruction algorithm from calibrated photographs based on the quantitative evaluation conducted by Computer Vision researchers. The MVS software has been used at numerous academic and industrial settings, including several visual effect companies, Industrial Light and Magic and Weta digital, for real film-production purposes, and at Google for Google Maps products. He is a pioneer in the automated 3D reconstruction and visualization techniques for indoor scenes, where one of his papers won the Best Student Paper Award at ECCV 2012. He also won the best paper award at 3DV 2013 for high fidelity geometric and photometric modeling from Internet photos. He received a prestigious NSF CAREER Award in 2015. Before joining Washington University, he was a software engineer at Google, and a research associate at the University of Washington. He earned his Ph.D. from the University of Illinois at Urbana-Champaign in 2008 and a Bachelor's degree from the University of Tokyo in 2001.