Centre de documentation
scientifique

Introducing the boundary-aware loss for deep image segmentation / Minh On Vu Ngoc (2021)  

Public Titre : Introducing the boundary-aware loss for deep image segmentation Type de document : Article/Communication Auteurs : Minh On Vu Ngoc, Auteur ; Yizi Chen , Auteur ; Nicolas Boutry, Auteur ; Joseph Chazalon, Auteur ; Edwin Carlinet, Auteur ; Jonathan Fabrizio, Auteur ; Clément Mallet , Auteur ; Thierry Géraud, Auteur Editeur : The British Machine Vision Association Press (BMVA) Année de publication : 2021 Projets : SODUCO / Perret, Julien Conférence : BMVC 2021, 32nd British Machine Vision Conference 22/11/2021 25/11/2021 online Royaume-Uni OA Proceedings Importance : 17 p. Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image [Termes IGN] apprentissage profond [Termes IGN] classification barycentrique [Termes IGN] segmentation d'image Résumé : (auteur) Most contemporary supervised image segmentation methods do not preserve the initial topology of the given input (like the closeness of the contours). One can generally remark that edge points have been inserted or removed when the binary prediction and the ground truth are compared. This can be critical when accurate localization of multiple interconnected objects is required. In this paper, we present a new loss function, called, Boundary-Aware loss (BALoss), based on the Minimum Barrier Distance (MBD) cut algorithm. It is able to locate what we call the leakage pixels and to encode the boundary information coming from the given ground truth. Thanks to this adapted loss, we are able to significantly refine the quality of the predicted boundaries during the learning procedure. Furthermore, our loss function is differentiable and can be applied to any kind of neural network used in image processing. We apply this loss function on the standard U-Net and DC U-Net on Electron Microscopy datasets. They are well-known to be challenging due to their high noise level and to the close or even connected objects covering the image space. Our segmentation performance, in terms of Variation of Information (VOI) and Adapted Rank Index (ARI), are very promising and lead to 15% better scores of VOI and 5% better scores of ARI than the state-of-the-art. The code of boundary-awareness loss is freely available at https://github.com/onvungocminh/MBD_BAL Numéro de notice : C2021-054 Affiliation des auteurs : UGE-LASTIG+Ext (2020- ) Thématique : IMAGERIE Nature : Communication nature-HAL : ComAvecCL&ActesPubliésIntl DOI : sans En ligne : https://www.bmvc2021-virtualconference.com/assets/papers/1546.pdf Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=99411

Leveraging class hierarchies with metric-guided prototype learning / Vivien Sainte Fare Garnot (2021)  

Public Titre : Leveraging class hierarchies with metric-guided prototype learning Type de document : Article/Communication Auteurs : Vivien Sainte Fare Garnot , Auteur ; Loïc Landrieu , Auteur Editeur : Ithaca [New York - Etats-Unis] : ArXiv - Université Cornell Année de publication : 2021 Projets : 1-Pas de projet / Perret, Julien Conférence : BMVC 2021, 32nd British Machine Vision Conference 22/11/2021 25/11/2021 online Royaume-Uni OA Proceedings Importance : 31 p. Note générale : bibliographie préprint déposé sur ArXiv Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Intelligence artificielle [Termes IGN] apprentissage automatique [Termes IGN] classification [Termes IGN] matrice d'erreur [Termes IGN] prototype [Termes IGN] segmentation sémantique Résumé : (auteur) Not all errors are created equal. This is especially true for many key machine learning applications. In the case of classification tasks, the severity of errors can be summarized under the form of a cost matrix, which assesses the gravity of confusing each pair of classes. When the target classes are organized into a hierarchical structure, this matrix defines a metric. We propose to integrate this metric in a new and versatile classification layer in order to model the disparity of errors. Our method relies on jointly learning a feature-extracting network and a set of class representations, or prototypes, which incorporate the error metric into their relative arrangement in the embedding space. Our approach allows for consistent improvement of the severity of the network's errors with regard to the cost matrix. Furthermore, when the induced metric contains insight on the data structure, our approach improves the overall precision as well. Experiments on four different public datasets -- from agricultural time series classification to depth image semantic segmentation -- validate our approach. Numéro de notice : C2021-027 Affiliation des auteurs : UGE-LASTIG (2020- ) Autre URL associée : vers ArXiv Thématique : IMAGERIE/INFORMATIQUE Nature : Poster nature-HAL : Poster-avec-CL DOI : 10.48550/arXiv.2007.03047 En ligne : https://www.bmvc2021-virtualconference.com/assets/papers/0084.pdf Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=98983