Pré-Publication, Document De Travail Année : 2025

Hierarchy-based fuzzy segmentation and marker learning layer: theory and algorithms

Résumé

Image segmentation is inherently challenging because it is often difficult to automatically identify the object of interest in an image. To mitigate this, human-provided markers can be incorporated into the segmentation process, greatly improving accuracy. However, human interaction is an expensive resource and methods that reduce effort in interactive segmentation are of great interest. In this work, we introduce a novel marker-based segmentation layer for deep neural networks, enabling end-to-end training of a marker creation network. Our training methodology includes a loss function with two main components: (i) segmentation loss using the new differentiable segmentation layer, and (ii) a set of regularization functions that ensure the generated markers have the desired shape properties. We show that by using the proposed method, the network can automatically generate markers that achieve effective segmentation and have desirable shape characteristics. We validate our results in the training dataset and in five unseen datasets.

Fichier principal
Vignette du fichier
2024_journal_learning_markers_gabriel_jmiv___Submitted_Version (2).pdf (16.21 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05040261 , version 1 (23-06-2025)

Licence

Identifiants

  • HAL Id : hal-05040261 , version 1

Citer

Gabriel Barbosa da Fonseca, Romain Negrel, Benjamin Perret, Jean Cousty, Silvio J.F. Guimarães. Hierarchy-based fuzzy segmentation and marker learning layer: theory and algorithms. 2025. ⟨hal-05040261⟩
161 Consultations
158 Téléchargements

Partager

  • More