Vous êtes ici : Accueil » Kiosque » Annonce


Mot de passe : 

Mot de passe oublié ?
Détails d'identification oubliés ?


7 novembre 2017

Deep learning for the detection and tracking of geophysical ocean structures

Catégorie : Stagiaire

CNN-based segmentation and tracking of physical structures in satellite-derived sea surface image time series.


Supervisor: Ronan Fablet, Cédric Herzet (ronan.fablet@imt-atlantique.fr)

Research team: IMT Atlantique, Lab-STICC, TOMS, Brest

Expected duration: 6 months

Contact person: ronan.fablet@imt-atlantique.fr

Scientific context and specific objective

Deep learning [1] has experienced tremendous growth in a few years in the field of artificial intelligence and computer vision. Initially exploited for classification and recognition problems, it has also become a reference framework for the resolution of signal and image processing problems: image synthesis, super-resolution, denoising, inpainting, segmentation...

The detection and tracking of physical structures (e.g., fronts, filaments, eddies...) in satellite-derived sea surface observations of the sea surface is a key issue for the characterization and understanding of the upper ocean dynamics [2,3,4]. Most approaches rely on rule-based algorithms [2,3] and only few studies have explored machine learning strategies [4]. The goal of this internship will be to develop and evaluate deep learning models, especially CNN-based (Convolutional Neural Net) segmentation models [5,6,7], for these detection and tracking tasks. Two specific objectives will be of particular interest:

Case-studies for real satellite-derived observation datasets [e.g., 3] will be considered. All experiments will be implemented under Python using dedicated libraries such as Keras and/or Tensorflow frameworks.

Keywords: neural networks, CNN, objection detection and tracking, sea surface dynamics, physical structures, multi-source satellite observations


The envisioned workplan involves three main aspects:


[1] Y. LeCun, Y. Bengio, and G. Hinton. Deep learning. Nature, 521(7553):436– 444, May 2015. 

[2] D. B. Chelton, M. G. Schlax, R. M. Samelson, and R. A. de Szoeke. Global observations of large oceanic eddies. Geophysical Research Letters, vol. 34, no. 15, 2007.

[3] E. Mason, A. Pascual, and J. C. McWilliams. A new sea surface height–based code for oceanic mesoscale eddy tracking. Journal of Atmospheric and Oceanic Technology, vol. 31, no. 5, pp. 1181–1188, 2014.

[4] M. D. Ashkezari, C. N. Hill, C. N. Follett, G. Forget, and M. J. Follows. Oceanic eddy detection and lifetime forecast using machine learning methods. Geophysical Research Letters, vol. 43, no. 23, 2016.

[5] Vijay Badrinarayanan, Alex Kendall and Roberto Cipolla "SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation. IEEE PAMI, 2017.

[6] N. Audebert, B. L. Saux, and S. Lefevre. Semantic segmentation of earth observation data using multimodal and multi-scale deep networks. arXiv preprint arXiv:1609.06846, 2016.

[7] O. Ronneberger, P. Fischer, and T. Brox. U-net: Convolutional networks for biomedical image segmentation. Int. Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, 2015, pp. 234–241.


Dans cette rubrique

(c) GdR 720 ISIS - CNRS - 2011-2018.