DS-Depth: Dynamic and Static Depth Estimation via a Fusion Cost Volume
- 15 August 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 34 (4) , 2564-2576
Abstract
Self-supervised monocular depth estimation methods typically rely on the reprojection error to capture geometric relationships between successive frames in static environments. However, this assumption does not hold in dynamic objects in scenarios, leading to errors during the view synthesis stage, such as feature mismatch and occlusion, which can significantly reduce the accuracy of the generated depth maps. To address this problem, we propose a novel dynamic cost volume that exploits residual optical flow to describe moving objects, improving incorrectly occluded regions in static cost volumes used in previous work. Nevertheless, the dynamic cost volume inevitably generates extra occlusions and noise, thus we alleviate this by designing a fusion module that makes static and dynamic cost volumes compensate for each other. In other words, occlusion from the static volume is refined by the dynamic volume, and incorrect information from the dynamic volume is eliminated by the static volume. Furthermore, we propose a pyramid distillation loss to reduce photometric error inaccuracy at low resolutions and an adaptive photometric error loss to alleviate the flow direction of the large gradient in the occlusion regions. We conducted extensive experiments on the KITTI and Cityscapes datasets, and the results demonstrate that our model outperforms previously published baselines for self-supervised monocular depth estimation.Keywords
All Related Versions
Funding Information
- Securing the Energy/Transport Interface (EP/X037401/1)
- U.K. Medical Research Council (MRC) Innovation Fellowship (MR/S003916/2)
- International Exchanges 2022 (IEC\NSFC\223523)
This publication has 71 references indexed in Scilit:
- Disentangling Object Motion and Occlusion for Unsupervised Multi-frame Monocular Depth2022
- EfficientPS: Efficient Panoptic SegmentationInternational Journal of Computer Vision, 2021
- Feature-Metric Loss for Self-supervised Learning of Depth and Egomotion2020
- Self-supervised Monocular Depth Estimation: Solving the Dynamic Object Problem by Semantic Guidance2020
- What Matters in Unsupervised Optical Flow2020
- Every Pixel Counts: Unsupervised Geometry Learning with Holistic 3D Motion Understanding2019
- MVSNet: Depth Inference for Unstructured Multi-view Stereo2018
- Learning Monocular Depth Estimation with Unsupervised Trinocular Assumptions2018
- Sparsity Invariant CNNs2017
- Unsupervised CNN for Single View Depth Estimation: Geometry to the Rescue2016