Dyna-DM: Dynamic Object-aware Self-supervised Monocular Depth Maps
2023
Abstract
Self-supervised monocular depth estimation has been a subject of intense study in recent years, because of its applications in robotics and autonomous driving. Much of the recent work focuses on improving depth estimation by increasing architecture complexity. This paper shows that state-of-the-art performance can also be achieved by improving the learning process rather than increasing model complexity. More specifically, we propose (i) disregarding small potentially dynamic objects when training, and (ii) employing an appearance-based approach to separately estimate object pose for truly dynamic objects. We demonstrate that these simplifications reduce GPU memory usage by 29% and result in qualitatively and quantitatively improved depth maps.
Citation
Kieran Saunders, George Vogiatzis and Luis J. Manso. “Dyna-DM: Dynamic Object-aware Self-supervised Monocular Depth Maps.” 2023.
BibTeX
@inproceedings{saunders2023,
title = {Dyna-DM: Dynamic Object-aware Self-supervised Monocular Depth Maps},
author = {Kieran Saunders and George Vogiatzis and Luis J. Manso},
pages = {10--16},
year = {2023},
doi = {10.1109/icarsc58346.2023.10129564},
}