EmNeF: Neural Fields for Embedded Variational Problems in Imaging

Danielle Bednarski*, Jan Lellmann

*Corresponding author for this work

Abstract

We propose a model-driven neural fields approach for solving variational problems. The approach can be applied to a variety of problems with convex, 1-homogeneous regularizer and arbitrary, possibly non-convex, data term. Our strategy is to embed the non-convex energy into a higher-dimensional space, reaching a convex primal-dual formulation. Instead of using classical gradient-descent based optimization algorithms, we propose training multiple fields representing the primal and dual variables in order to solve the problem.
Original languageEnglish
Title of host publicationScale Space and Variational Methods in Computer Vision : 9th International Conference, SSVM 2023, Santa Margherita di Pula, Italy, May 21-25, 2023, Proceedings
Number of pages12
PublisherSpringer, Cham
Publication date10.05.2023
Pages 137 - 148
ISBN (Print)9783031319747
DOIs
Publication statusPublished - 10.05.2023

Cite this