Skip to Main content Skip to Navigation
Conference papers

Co-segmentation non-supervisée d'images utilisant les distances de Sinkhorn

Julien Rabin 1, * Nicolas Papadakis 2
* Corresponding author
1 Equipe Image - Laboratoire GREYC - UMR6072
GREYC - Groupe de Recherche en Informatique, Image, Automatique et Instrumentation de Caen
Abstract : In this work, a convex and robust formulation of the unsupervised co-segmentation problem is introduced for pair of images. The proposed model relies on the optimal transport theory to asset the statistical similarity of the segmented regions’ features (color histograms in this work). The optimal transport cost is approximated by Sinkhorn distance to reduce the optimization complexity. A primal-dual algorithm is used to solve the problem efficiently, without making use of sub-iterative routines.
Document type :
Conference papers
Complete list of metadatas

Cited literature [8 references]  Display  Hide  Download
Contributor : Julien Rabin <>
Submitted on : Thursday, September 17, 2015 - 1:11:53 PM
Last modification on : Monday, July 22, 2019 - 11:00:24 AM
Document(s) archivé(s) le : Tuesday, December 29, 2015 - 7:44:43 AM


Files produced by the author(s)


Distributed under a Creative Commons Attribution - NoDerivatives 4.0 International License


  • HAL Id : hal-01200862, version 1


Julien Rabin, Nicolas Papadakis. Co-segmentation non-supervisée d'images utilisant les distances de Sinkhorn. Colloque GRETSI 2015, Sep 2015, Lyon, France. ⟨hal-01200862⟩



Record views


Files downloads