arXiv Analytics

Sign in

arXiv:2104.03308 [cs.CV]AbstractReferencesReviewsResources

Warp Consistency for Unsupervised Learning of Dense Correspondences

Prune Truong, Martin Danelljan, Fisher Yu, Luc Van Gool

Published 2021-04-07Version 1

The key challenge in learning dense correspondences lies in the lack of ground-truth matches for real image pairs. While photometric consistency losses provide unsupervised alternatives, they struggle with large appearance changes, which are ubiquitous in geometric and semantic matching tasks. Moreover, methods relying on synthetic training pairs often suffer from poor generalisation to real data. We propose Warp Consistency, an unsupervised learning objective for dense correspondence regression. Our objective is effective even in settings with large appearance and view-point changes. Given a pair of real images, we first construct an image triplet by applying a randomly sampled warp to one of the original images. We derive and analyze all flow-consistency constraints arising between the triplet. From our observations and empirical results, we design a general unsupervised objective employing two of the derived constraints. We validate our warp consistency loss by training three recent dense correspondence networks for the geometric and semantic matching tasks. Our approach sets a new state-of-the-art on several challenging benchmarks, including MegaDepth, RobotCar and TSS. Code and models will be released at

Comments: code:
Categories: cs.CV
Related articles: Most relevant | Search more
arXiv:1803.01260 [cs.CV] (Published 2018-03-03)
Unsupervised Learning of Face Representations
arXiv:1811.02804 [cs.CV] (Published 2018-11-07)
Image Smoothing via Unsupervised Learning
arXiv:1511.04166 [cs.CV] (Published 2015-11-13)
Unsupervised Learning of Edges