Saved in:
Bibliographic Details
Main Authors: Gonzalez-Alvarado, Daniel, Schlindwein, Fabio, Cassel, Jonas, Steingruber, Laura, Petra, Stefania, Schnörr, Christoph
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2504.13024
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908332875317248
author Gonzalez-Alvarado, Daniel
Schlindwein, Fabio
Cassel, Jonas
Steingruber, Laura
Petra, Stefania
Schnörr, Christoph
author_facet Gonzalez-Alvarado, Daniel
Schlindwein, Fabio
Cassel, Jonas
Steingruber, Laura
Petra, Stefania
Schnörr, Christoph
contents This paper introduces patch assignment flows for metric data labeling on graphs. Labelings are determined by regularizing initial local labelings through the dynamic interaction of both labels and label assignments across the graph, entirely encoded by a dictionary of competing labeled patches and mediated by patch assignment variables. Maximal consistency of patch assignments is achieved by geometric numerical integration of a Riemannian ascent flow, as critical point of a Lagrangian action functional. Experiments illustrate properties of the approach, including uncertainty quantification of label assignments.
format Preprint
id arxiv_https___arxiv_org_abs_2504_13024
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Riemannian Patch Assignment Gradient Flows
Gonzalez-Alvarado, Daniel
Schlindwein, Fabio
Cassel, Jonas
Steingruber, Laura
Petra, Stefania
Schnörr, Christoph
Computer Vision and Pattern Recognition
This paper introduces patch assignment flows for metric data labeling on graphs. Labelings are determined by regularizing initial local labelings through the dynamic interaction of both labels and label assignments across the graph, entirely encoded by a dictionary of competing labeled patches and mediated by patch assignment variables. Maximal consistency of patch assignments is achieved by geometric numerical integration of a Riemannian ascent flow, as critical point of a Lagrangian action functional. Experiments illustrate properties of the approach, including uncertainty quantification of label assignments.
title Riemannian Patch Assignment Gradient Flows
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.13024