Learnable Cost Volume Using the Cayley Representation

Taihong Xiao, Jinwei Yuan, Deqing Sun, Qifei Wang, Xin Yu Zhang, Kehan Xu, Ming Hsuan Yang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Cost volume is an essential component of recent deep models for optical flow estimation and is usually constructed by calculating the inner product between two feature vectors. However, the standard inner product in the commonly-used cost volume may limit the representation capacity of flow models because it neglects the correlation among different channel dimensions and weighs each dimension equally. To address this issue, we propose a learnable cost volume (LCV) using an elliptical inner product, which generalizes the standard inner product by a positive definite kernel matrix. To guarantee its positive definiteness, we perform spectral decomposition on the kernel matrix and re-parameterize it via the Cayley representation. The proposed LCV is a lightweight module and can be easily plugged into existing models to replace the vanilla cost volume. Experimental results show that the LCV module not only improves the accuracy of state-of-the-art models on standard benchmarks, but also promotes their robustness against illumination change, noises, and adversarial perturbations of the input signals.

Original languageEnglish
Title of host publicationComputer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings
EditorsAndrea Vedaldi, Horst Bischof, Thomas Brox, Jan-Michael Frahm
PublisherSpringer Science and Business Media Deutschland GmbH
Pages483-499
Number of pages17
ISBN (Print)9783030585440
DOIs
Publication statusPublished - 2020
Event16th European Conference on Computer Vision, ECCV 2020 - Glasgow, United Kingdom
Duration: 2020 Aug 232020 Aug 28

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12354 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference16th European Conference on Computer Vision, ECCV 2020
CountryUnited Kingdom
CityGlasgow
Period20/8/2320/8/28

Bibliographical note

Funding Information:
Acknowledgments. This work is supported in part by NSF CAREER Grant 1149783. We also thank Pengpeng Liu and Jingfeng Wu for kind help.

Publisher Copyright:
© 2020, Springer Nature Switzerland AG.

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Learnable Cost Volume Using the Cayley Representation'. Together they form a unique fingerprint.

Cite this