SC-DepthV3: Robust Self-Supervised Monocular Depth Estimation for Dynamic Scenes
IEEE Transactions on Pattern Analysis and Machine Intelligence · 2024-01 · 46 (1) · 497-508
DOI: 10.1109/tpami.2023.3322549 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024-01. Verfasst von Libo Sun und 5 weiteren Autor:innen. Die Arbeit wurde bisher 108-mal zitiert (Crossref) und nennt 75 Literaturangaben. Gefördert von: National Key R&D Program of China, Australian Centre of Excellence for Robotic Vision, ARC Laureate Fellowship.
Abstract (Originalsprache)
Self-supervised monocular depth estimation has shown impressive results in static scenes. It relies on the multi-view consistency assumption for training networks, however, that is violated in dynamic object regions and occlusions. Consequently, existing methods show poor accuracy in dynamic scenes, and the estimated depth map is blurred at object boundaries because they are usually occluded in other training views. In this paper, we propose SC-DepthV3 for addressing the challenges. Specifically, we introduce an external pretrained monocular depth estimation model for generating single-image depth prior, namely pseudo-depth, based on which we propose novel losses to boost self-supervised training. As a result, our model can predict sharp and accurate depth maps, even when training from monocular videos of highly dynamic scenes. We demonstrate the significantly superior performance of our method over previous methods on six challenging datasets, and we provide detailed ablation studies for the proposed terms.
Förderung
- National Key R&D Program of China
- Australian Centre of Excellence for Robotic Vision
- ARC Laureate Fellowship
Zitieren
APA
Sun, L., Bian, J. W., Zhan, H., Yin, W., Reid, I., & Shen, C. (2024). SC-DepthV3: Robust Self-Supervised Monocular Depth Estimation for Dynamic Scenes. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(1), 497-508. https://doi.org/10.1109/tpami.2023.3322549
Vancouver
Sun L, Bian JW, Zhan H, Yin W, Reid I, Shen C. SC-DepthV3: Robust Self-Supervised Monocular Depth Estimation for Dynamic Scenes. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2024;46(1):497-508. doi:10.1109/tpami.2023.3322549
BibTeX
@article{Sun2024,
author = {Sun, Libo and Bian, Jia-Wang and Zhan, Huangying and Yin, Wei and Reid, Ian and Shen, Chunhua},
title = {SC-DepthV3: Robust Self-Supervised Monocular Depth Estimation for Dynamic Scenes},
journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
year = {2024},
volume = {46},
number = {1},
pages = {497--508},
doi = {10.1109/tpami.2023.3322549}
}
EUVIMED