Computer Vision News - June 2019

quickly before moving on to a more robust, accurate localization. Vassileios expands: “ I think this is also important: to make new methods that combine the two. It is also a very generic problem. You have to be robust to different things, but not to get lost in scenes that might look similar. Imagine you have a huge city. You have so many places that look exactly like each other. You have streets with repeating patterns and things like that. How do you localize in different levels of hierarchy of precision? ” In recent work, they have found that it helps to integrate some sort of higher level scene understanding, for example, sample semantic segmentation, into the localization process, similar to what humans do. One problem that they ran into is that those semantic segmentation algorithms are not necessarily robust against all types of changes. In that sense, the datasets are a challenge. Organizers and speakers at the upcoming workshop are experts from both industry and academia. The blend of expertise at the workshop will give an interesting perspective from different communities which are often disconnected. Jan-Michael Frahm , the head of the 3D computer vision group at the University of North Carolina at Chapel Hill , will share his experience researching a variety of topics on the intersection of computer vision, computer graphics, robotics. Srikumar Ramalingam , associate professor in the school of computing at the University of Utah , will also join, sharing his research on computer vision, machine learning, robotics, and autonomous driving. Niko Sünderhauf , senior lecturer at the Queensland University of Technology (QUT) in Brisbane, Australia , will share his insights on robotic vision, the intersection of robotics, computer vision, and machine learning. Also set to speak: Simon Lynen , a tech lead/manager at Google Zurich , who works on large-scale localization from an industrial point of view. He can provide interesting feedback about the problems they run into that others may not have confronted. The Long-Term Visual Localization under Changing Conditions will take place on Monday, June 17th from 1:30 PM to 6 PM at CVPR 2019 in Long Beach, California. Computer Vision News will be there to publish for the 4th consecutive year the CVPR Daily magazine! Pauline Luc 31 Computer Vision News Challenge The website of the workshop and the Impressive list of organizers are at https://sites.google.com/view/ltvl2019/ Photos in this article belong to their respective owners and are shared under CC BY-NC-SA 4.0 Long-Term Visual Localization u der …