Practical Deep Feature-Based Visual-Inertial Odometry

Charles Hamesse, Charles Hamesse, Michiel Vlaminck, Hiep Luong, Rob Haelterman

2024

Abstract

We present a hybrid visual-inertial odometry system that relies on a state-of-the-art deep feature matching front-end and a traditional visual-inertial optimization back-end. More precisely, we develop a fully-fledged feature tracker based on the recent SuperPoint and LightGlue neural networks, that can be plugged directly to the estimation back-end of VINS-Mono. By default, this feature tracker returns extremely abundant matches. To bound the computational complexity of the back-end optimization, limiting the number of used matches is desirable. Therefore, we explore various methods to filter the matches while maintaining a high visual-inertial odometry performance. We run extensive tests on the EuRoC machine hall and Vicon room datasets, showing that our system achieves state-of-the-art odometry performance according relative pose errors.

Paper Citation

in Harvard Style

Hamesse C., Vlaminck M., Luong H. and Haelterman R. (2024). Practical Deep Feature-Based Visual-Inertial Odometry. In Proceedings of the 13th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM; ISBN 978-989-758-684-2, SciTePress, pages 240-247. DOI: 10.5220/0012320200003654

in Bibtex Style

@conference{icpram24,
author={Charles Hamesse and Michiel Vlaminck and Hiep Luong and Rob Haelterman},
title={Practical Deep Feature-Based Visual-Inertial Odometry},
booktitle={Proceedings of the 13th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM},
year={2024},
pages={240-247},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012320200003654},
isbn={978-989-758-684-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 13th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM
TI - Practical Deep Feature-Based Visual-Inertial Odometry
SN - 978-989-758-684-2
AU - Hamesse C.
AU - Vlaminck M.
AU - Luong H.
AU - Haelterman R.
PY - 2024
SP - 240
EP - 247
DO - 10.5220/0012320200003654
PB - SciTePress