Accéder directement au contenu Accéder directement à la navigation
Communication dans un congrès

Transfer learning on CNN architectures for ship classification on SAR images

Abdelmalek Toumi 1, 2 Jean-Christophe Cexus 1, 2 Ali Khenchaf 3, 2 Antoine Tartivel 2
1 Lab-STICC_ENSTAB_CID_TOMS
Lab-STICC - Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance
2 Pôle STIC_REMS
ENSTA Bretagne - École Nationale Supérieure de Techniques Avancées Bretagne
3 Lab-STICC_ENSTAB_MOM_PIM
Lab-STICC - Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance
Abstract : Synthetic-aperture radar (SAR) imagery has great potential for maritime surveillance with its global coverage as well as its weather independence. In order to leverage this potential, machine learning can be used to automatically process large amounts of data for different goals. This work focuses on the use of deep learning algorithms for ship classification. Particularly, the potential of transfer learning applied to convolutional neural networks (CNNs) is assessed in this context. This is especially relevant for tasks like this one, where there is no huge labelled dataset available for training. The aim is thus to see how to leverage knowledge from models pre-trained on other tasks (source tasks) and use them for ship classification (target task).
Liste complète des métadonnées

https://hal-ensta-bretagne.archives-ouvertes.fr/hal-03109596
Contributeur : Jean-Christophe Cexus <>
Soumis le : mercredi 13 janvier 2021 - 19:42:01
Dernière modification le : mercredi 21 avril 2021 - 11:40:05

Identifiants

  • HAL Id : hal-03109596, version 1

Citation

Abdelmalek Toumi, Jean-Christophe Cexus, Ali Khenchaf, Antoine Tartivel. Transfer learning on CNN architectures for ship classification on SAR images. Sea Tech Week - Session Remote Sensing, Oct 2020, Brest, France. ⟨hal-03109596⟩

Partager

Métriques

Consultations de la notice

53