Laboratoire de Génie Informatique et d’Automatique de l’Artois

Ph.D. thesis of Pauline MINARY

Evidential calibration and fusion of multiple classifiers: application to face blurring

Starting date: 1 December 2014
Funding: CIFRE grant in partnership with the SNCF
Keywords: Video sequences, blurring faces, face detectors, information fusion, belief functions
Advising:

To ensure safely trains departures, especially in a high-density environment, the SNCF offers the driver of the train a camera system allowing him/her to see all of his train within monitors displaying images of the platform.

To allow the SNCF to use these videos for other purposes, such as studying for example moving users in a station, it is necessary to respect the right of publicity and then to blur people faces on images.

The objective of this thesis is then to realize a robust face detection system in order to blur faces. This system will in particular be confronted to uncontrolled conditions (weather, exterior, interior, ...) in which operate the cameras.

Involved research themes:

Involved application areas:

Partners

LGI2A

Université d'Artois
France

SNCF Réseau

France

Defense

Defense took place the 08/12/2017 am31 10:00 Salle Prestige, FSA, Béthune

Jury:

  • Rapporteur Thierry DENOEUX Université de Technologie de Compiègne
  • Rapporteur Michèle ROMBAUT Université Grenoble Alpes
  • Examinateur Sylvie LE HEGARAT-MASCLE Polytech Paris sud
  • Examinateur Olivier COLOT Université Lille 1
  • Président Franck LUTHON Université Pau et Pays de l'Adour
  • Co-supervisor Frédéric PICHON Université d'Artois
  • Co-directeur David MERCIER Université d'Artois
  • Directeur Eric LEFEVRE Université d'Artois
  • Invité Benjamin DROIT SNCF RESEAU