Le Dr. Christopher Zach de Toshiba Research Cambridge donnera un séminaire intitulé "A Tour of Optimization Methods for 3D Computer Vision", mardi 9 mai, de 11h à 12h,
dans l'Amphithéâtre Jean-Paul Dom du laboratoire IMS.
Dr Christopher Zach from Toshiba Research Cambridge will be visiting our institution on the 9th and 10th of May.
On this occasion, he will be giving a presentation on optimization methods for 3D computer vision. The presentation will take place on Tuesday, May the 9th at 11am at the IMS laboratory (amphithéâtre Jean-Paul Dom).
A Tour of Optimization Methods for 3D Computer Vision
In this talk I give an overview of my research activities in various computer vision labs, most notably ETH Zurich, MSR Cambridge and Toshiba Research
Europe. One major motivation and application behind my research is 3D modelling from image and video data. I have worked on a broad spectrum of
topics ranging from structure-and-motion computation to volumetric and convex approaches for dense surface modelling. In this presentation I will provide an overview of how a range of novel optimization approaches helps to solve several computer vision problems such as robust estimation, dense 3D modelling, and combining 3D reconstruction with semantic cues. The underlying mathematical tools to address these problems are convex formulations and relaxations, efficient inference in graphical models, and statistical
reasoning. I will in particular highlight two topics in my presentation: a simple method for robust non-linear least squares optimization, and a convex relaxation framework for piecewise convex functions.