Talk by Martin Oswald (ETH Zürich) on “Image-based 3D Scene Understanding”
Time: Wednesday 21 October 2020, 11:00 am
Online via meeting tool
Abstract:
Understanding a 3D scene from images is an effortless task for humans, but still highly challenging for machines. Nevertheless there are numerous useful applications like autonomous driving, robotics, human computer interaction, and 3D content creation for games, movies and mixed reality. Depending on the task at hand, we are interested in estimating the 3D shape of objects, their semantic meaning, object motion, as well as appearance and light information.
In my talk, I will give an overview of my research on image-based 3D scene understanding and discuss aspects like scene representations, representation learning, semantic understanding, as well as scalability and real-time processing considerations. I will show on several examples how machine learning, optimization techniques and problem tailored shape representations help to make 3D scene understanding methods more accurate, scalable and robust.
Bio:
Martin R. Oswald is a senior researcher, lecturer and deputy lab director in the Computer Vision and Geometry lab at ETH Zurich, headed by Prof. Marc Pollefeys. Starting in February 2021 he will be a Tenure Track Assistant Professor for 3D Computer Vision at Vienna University of Technology (TU Wien).
He obtained his PhD in Computer Vision from TU Munich under the supervision of Prof. Daniel Cremers in 2015. Previously, he received a masters degree in civil engineering from the University of Technology Federico Santa Maria in Valparaiso, Chile in 2008 supported by a DAAD fellowship program. Before that he studied computer science at TU Dresden and the University of Technology Sydney and obtained a master degree (Diplom) from TU Dresden in 2007. His works received several awards including 3DV 2019 best paper honorable mention award, the DAGM 2009 Paper Award and the ACCV Honorable Mention Award in 2010.
