AR on handheld, monocular, "through-the-camera" platforms such as mobile phones is a challenging task. While traditional, geometry based approaches provide useful data in certain scenarios, for truly immersive experiences we need to leverage the prior knowledge encapsulated in learned CNNs. In this talk I will discuss the capabilities and limitations of such traditional methods, the need for CNN-based solutions, and the challenges to training accurate and efficient CNNs on this task. I will describe our recent work on implicit, 3D representations for AR, with applications in novel view synthesis, scene reconstruction and arbitrary object manipulation. Finally, I will present a project opportunity, to learn such representations from a dataset of single images.
Oliver Woodford is a Lead Research Scientist in the Creative Vision Group at Snap Research in Los Angeles. His research focuses on statistical models and optimization, with applications in geometry and low-level vision. He obtained a DPhil from Oxford University in 2009, supervised by Andrew Fitzgibbon, Phil Torr and Ian Reid, for work on novel view synthesis and stereo, receiving a Best Paper Award at CVPR 2008 for the latter. Prior to joining Snap, he worked at Toshiba Research, where he received a prestigious Toshiba Research & Development Achievement Award in 2012, and mobile app startup Seene, winner of several tech awards including The Tech Expo Champion 2015.