MERLIN-Seg: self-supervised despeckling for label-efficient semantic segmentation - Equipe Image, Modélisation, Analyse, GEométrie, Synthèse Access content directly
Journal Articles Computer Vision and Image Understanding Year : 2024

MERLIN-Seg: self-supervised despeckling for label-efficient semantic segmentation

Abstract

Remote sensing satellites acquire a continuous stream of data on a daily basis. As most of those data are unlabeled, the development of algorithms requiring weak supervision is of paramount importance. In this paper, we show that the need for annotation for Synthetic Aperture Radar data can be reduced by coupling a despeckling task (self-supervised) and a segmentation task (supervised). The proposed self-supervised learning framework, called MERLIN-Seg, has been trained for building footprint extraction and achieves favorable performances even with 1% of annotated data. We show that conditioning the network on despeckling without labels is beneficial for supervised segmentation. Our experiments demonstrate that the joint training of the two tasks achieves better performances than a vanilla segmentation network in terms of IoU, F1 score, and accuracy on both simulated and real SAR images.
Fichier principal
Vignette du fichier
CVIU_MERLIN_Seg.pdf (5.1 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-04163624 , version 1 (17-07-2023)
hal-04163624 , version 2 (12-02-2024)

Identifiers

Cite

Emanuele Dalsasso, Clément Rambour, Nicolas Trouvé, Nicolas Thome. MERLIN-Seg: self-supervised despeckling for label-efficient semantic segmentation. Computer Vision and Image Understanding, 2024, 241, ⟨10.1016/j.cviu.2024.103940⟩. ⟨hal-04163624v2⟩
228 View
170 Download

Altmetric

Share

Gmail Facebook X LinkedIn More