On Power Jaccard Losses for Semantic Segmentation

David Duque-Arias, Santiago Velasco-Forero, Jean-Emmanuel Deschaud, François Goulette, Andres Serna, Etienne Decencière, Beatriz Marcotegui

2021

Abstract

In this work, a new generalized loss function is proposed called power Jaccard to perform semantic segmentation tasks. It is compared with classical loss functions in different scenarios, including gray level and color image segmentation, as well as 3D point cloud segmentation. The results show improved performance, stability and convergence. We made available the code with our proposal with a demonstrative example.

Paper Citation

in Harvard Style

Duque-Arias D., Velasco-Forero S., Deschaud J., Goulette F., Serna A., Decencière E. and Marcotegui B. (2021). On Power Jaccard Losses for Semantic Segmentation. In Proceedings of the 16th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2021) - Volume 5: VISAPP; ISBN 978-989-758-488-6, SciTePress, pages 561-568. DOI: 10.5220/0010304005610568

in Bibtex Style

@conference{visapp21,
author={David Duque-Arias and Santiago Velasco-Forero and Jean-Emmanuel Deschaud and François Goulette and Andres Serna and Etienne Decencière and Beatriz Marcotegui},
title={On Power Jaccard Losses for Semantic Segmentation},
booktitle={Proceedings of the 16th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2021) - Volume 5: VISAPP},
year={2021},
pages={561-568},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010304005610568},
isbn={978-989-758-488-6},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 16th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2021) - Volume 5: VISAPP
TI - On Power Jaccard Losses for Semantic Segmentation
SN - 978-989-758-488-6
AU - Duque-Arias D.
AU - Velasco-Forero S.
AU - Deschaud J.
AU - Goulette F.
AU - Serna A.
AU - Decencière E.
AU - Marcotegui B.
PY - 2021
SP - 561
EP - 568
DO - 10.5220/0010304005610568
PB - SciTePress