Sublabel-Accurate Multilabeling Meets Product Label Spaces
Ye Z, Haefner B, Quéau Y, Möllenhoff T, Cremers D (2021)
Publication Type: Conference contribution
Publication year: 2021
Journal
Publisher: Springer Science and Business Media Deutschland GmbH
Book Volume: 13024 LNCS
Pages Range: 3-17
Conference Proceedings Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Event location: Virtual, Online
ISBN: 9783030926588
DOI: 10.1007/978-3-030-92659-5_1
Abstract
Functional lifting methods are a promising approach to determine optimal or near-optimal solutions to difficult nonconvex variational problems. Yet, they come with increased memory demands, limiting their practicability. To overcome this drawback, this paper presents a combination of two approaches designed to make liftings more scalable, namely product-space relaxations and sublabel-accurate discretizations. Our main contribution is a simple way to solve the resulting semi-infinite optimization problem with a sampling strategy. We show that despite its simplicity, our approach significantly outperforms baseline methods, in the sense that it finds solutions with lower energies given the same amount of memory. We demonstrate our empirical findings on the nonconvex optical flow and manifold-valued denoising problems.
Involved external institutions
Groupe de Recherche en Informatique, Image, Automatique et Instrumentation de Caen (GREYC)
France (FR)
Technische Universität München (TUM)
Germany (DE)
Riken Institute of Physical and Chemical Research / 理化学研究所
Japan (JP)
France (FR)
Technische Universität München (TUM)
Germany (DE)
Riken Institute of Physical and Chemical Research / 理化学研究所
Japan (JP)
How to cite
APA:
Ye, Z., Haefner, B., Quéau, Y., Möllenhoff, T., & Cremers, D. (2021). Sublabel-Accurate Multilabeling Meets Product Label Spaces. In Christian Bauckhage, Juergen Gall, Alexander Schwing (Eds.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 3-17). Virtual, Online: Springer Science and Business Media Deutschland GmbH.
MLA:
Ye, Zhenzhang, et al. "Sublabel-Accurate Multilabeling Meets Product Label Spaces." Proceedings of the 43rd DAGM German Conference on Pattern Recognition, DAGM GCPR 2021, Virtual, Online Ed. Christian Bauckhage, Juergen Gall, Alexander Schwing, Springer Science and Business Media Deutschland GmbH, 2021. 3-17.
BibTeX: Download