Constant Velocity Constraints for Self-Supervised Monocular Depth Estimation

Zhou, Hang, Greenwood, David, Taylor, Sarah and Gong, Han (2020) Constant Velocity Constraints for Self-Supervised Monocular Depth Estimation. In: CVMP '20: European Conference on Visual Media Production. Association for Computing Machinery (ACM), pp. 1-8. ISBN 9781450381987

[img]
Preview
PDF (Accepted_Manuscript) - Accepted Version
Download (7MB) | Preview

Abstract

We present a new method for self-supervised monocular depth estimation. Contemporary monocular depth estimation methods use a triplet of consecutive video frames to estimate the central depth image. We make the assumption that the ego-centric view progresses linearly in the scene, based on the kinematic and physical properties of the camera. During the training phase, we can exploit this assumption to create a depth estimation for each image in the triplet. We then apply a new geometry constraint that supports novel synthetic views, thus providing a strong supervisory signal. Our contribution is simple to implement, requires no additional trainable parameter, and produces competitive results when compared with other state-of-the-art methods on the popular KITTI corpus.

Item Type: Book Section
Uncontrolled Keywords: deep learning,monocular depth estimation,self-supervised learning,computer vision and pattern recognition,computer graphics and computer-aided design ,/dk/atira/pure/subjectarea/asjc/1700/1707
Faculty \ School: Faculty of Science > School of Computing Sciences
Related URLs:
Depositing User: LivePure Connector
Date Deposited: 12 Feb 2021 00:36
Last Modified: 11 Jul 2021 23:48
URI: https://ueaeprints.uea.ac.uk/id/eprint/79226
DOI: 10.1145/3429341.3429355

Actions (login required)

View Item View Item