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.
|Title of host publication||Scale Space and Variational Methods in Computer Vision : 9th International Conference, SSVM 2023, Santa Margherita di Pula, Italy, May 21-25, 2023, Proceedings|
|Pages||137 - 148|
|Publication status||Published - 10.05.2023|