Digging Into Self-Supervised Monocular Depth Estimation

Clement Godard, Oisin Mac Aodha, Michael Firman, Gabriel J. Brostow; The IEEE International Conference on Computer Vision (ICCV), 2019, pp. 3828-3838

Abstract


Per-pixel ground-truth depth data is challenging to acquire at scale. To overcome this limitation, self-supervised learning has emerged as a promising alternative for training models to perform monocular depth estimation. In this paper, we propose a set of improvements, which together result in both quantitatively and qualitatively improved depth maps compared to competing self-supervised methods. Research on self-supervised monocular training usually explores increasingly complex architectures, loss functions, and image formation models, all of which have recently helped to close the gap with fully-supervised methods. We show that a surprisingly simple model, and associated design choices, lead to superior predictions. In particular, we propose (i) a minimum reprojection loss, designed to robustly handle occlusions, (ii) a full-resolution multi-scale sampling method that reduces visual artifacts, and (iii) an auto-masking loss to ignore training pixels that violate camera motion assumptions. We demonstrate the effectiveness of each component in isolation, and show high quality, state-of-the-art results on the KITTI benchmark.

Related Material


[pdf]
[bibtex]
@InProceedings{Godard_2019_ICCV,
author = {Godard, Clement and Mac Aodha, Oisin and Firman, Michael and Brostow, Gabriel J.},
title = {Digging Into Self-Supervised Monocular Depth Estimation},
booktitle = {The IEEE International Conference on Computer Vision (ICCV)},
month = {October},
year = {2019}
}