%0 Conference Proceedings %T Improving constrained bundle adjustment through semantic scene labeling %+ Département Intelligence Ambiante et Systèmes Interactifs (DIASI) %+ Institut Pascal (IP) %A Salehi, A. %A Gay-Bellile, V. %A Bourgeois, S. %A Chausse, F. %Z Conference of 14th European Conference on Computer Vision, ECCV 2016 ; Conference Date: 8 October 2016 Through 16 October 2016; Conference Code:184029 %< avec comité de lecture %B ECCV 2016: Computer Vision – ECCV 2016 Workshops %C Amsterdam, Netherlands %3 Computer Vision – ECCV 2016 Workshops. ECCV 2016. Lecture Notes in Computer Science %V 9915 %P 133-142 %8 2016-10-08 %D 2016 %R 10.1007/978-3-319-49409-8_13 %K Learning systems %K Semantics %K Bundle adjustments %K City scale %K Constrained bundle adjustments %K Deep learning %K Learning process %K SLAM %K VSLAM %K Computer vision %Z Computer Science [cs]Conference papers %X There is no doubt that SLAM and deep learning methods can benefit from each other. Most recent approaches to coupling those two subjects, however, either use SLAM to improve the learning process, or tend to ignore the geometric solutions that are currently used by SLAM systems. In this work, we focus on improving city-scale SLAM through the use of deep learning. More precisely, we propose to use CNNbased scene labeling to geometrically constrain bundle adjustment. Our experiments indicate a considerable increase in robustness and precision. %G English %L cea-01813719 %U https://cea.hal.science/cea-01813719 %~ CEA %~ PRES_CLERMONT %~ CNRS %~ UNIV-BPCLERMONT %~ INSTITUT_PASCAL %~ DRT %~ CEA-UPSAY %~ UNIV-PARIS-SACLAY %~ CEA-UPSAY-SACLAY %~ LIST %~ GS-ENGINEERING %~ GS-COMPUTER-SCIENCE