Annonce en lien avec l’Action/le Réseau : aucun

Laboratoire/Entreprise : Oceanography Laboratory of Villefranche (LOV, UPMC-CNRS, Villefranche-sur-mer), French Riviera
Durée : 12 months, with an opportunity of extension
Contact : marc.picheral@obs-vlfr.fr and copy to antoine.manzanera@ensta-paristech.fr
Date limite de publication : 2016-09-10

Contexte :
The European H2020 BRIDGES project aims at developing a technology for deep-underwater glider(2400m and 6000m) and to propose sensors adapted to different uses of the vector. A smart imaging sensor named Octopus is being developed under the responsibility of the Oceanography Laboratory of Villefranche (LOV, UPMC-CNRS). It will detect, count and characterize all the objects larger than 100 microns. It will automatically recognize large objects (plankton, aggregates) with a minimal energy consumption increase. This knowledge should give the glider the opportunity to adapt its navigation in order to comply with its predefined mission.

Sujet :
OBJECTIVES OF THE POST-DOC

The development of machine learning methods using deep architectures while minimizing the resources required in the prediction step, thus taking into account the strong energy constraint and very limited space imposed by the glider system, Machine learning (using deep architecture) with contextual information such as metadata,
The selection of the hardware and the implementation of the proposed prediction method(s) on the Octopus smart sensor,
The study of Transfer Learning to evaluate the use of architectures learned on the plankton images of the Kaggle competition 3 to other plankton imaging systems with various characteristic (the high resolution ZooScan system, images acquired by the UVP5, FlowCam images…).

Profil du candidat :
A PhD in sciences with a strong knowledge of image processing and deep learning techniques. Solid algorithmic skills with a taste for efficient implementations and embedded computing would be particularly appreciated.

Formation et compétences requises :
Solid algorithmic skills with a taste for efficient implementations and embedded computing would be particularly appreciated.

Adresse d’emploi :
The future post-doctoral fellow will be based in Villefranche-sur-mer (Côte d’azur, France) in the Oceanography Laboratory of Villefranche under the responsibility of Marc Picheral, in charge of the development of the Octopus smart sensor. Scientific supervision
will be carried out in partnership with Antoine Manzanera from ENSTA, Paris. The post-doc funding is for 12 months, with an opportunity of extension.

Around 48 k€ gross yearly (to be adjusted depending on the candidate experience)

Send your CV, references and a letter in support of application to marc.picheral@obs-vlfr.fr and copy to antoine.manzanera@ensta-paristech.fr

Document attaché : post-doc_h2020_plankton_recognition.pdf