Vous êtes ici : Accueil » Kiosque » Annonce


Mot de passe : 

Mot de passe oublié ?
Détails d'identification oubliés ?


15 juin 2017

PhD position : Deep learning for ski lift video analysis

Catégorie : Doctorant


Safety in ski lifts is a major concern for ski resort operators. To prevent the risk of accident, it is necessary to detect dangerous situations when boarding or exiting vehicles (seats). The main goal of MIVAO project is to develop a computer vision system capable of analyzing videos filming the boarding scenes. This cutting edge project is led by the BlueCime company in partnership with Hubert Curien Laboratory, Saint-Étienne (France), Gipsa Lab, Grenoble and an increasing number of ski resorts in the Alpes mountains. The PhD candidate will join the working team of Mivao composed of three PhD students, one post-doc and more than 10 researchers and engineer. He will be hosted in Hubert Curien Laboratory in Saint-Étienne.

Scientific objectives

The objective of this thesis is to propose new computer vision methods to analyze the boarding scene and detect objects, persons attributes and situations that will help improving chairlift safety. These methods will be based on deep convolutional networks and will have to adapt automatically according to the context and weather conditions. The main issues to address will be:

  1. Design of a specific network to detect relevant elements of the scene such as the bodyguard, persons, or skis.
  2. Study of transfer learning methods to provide automatic adaptation to the context.
  3. Proposal of a model to predict anomalies according to the spatial disposition of the detected elements and their temporal evolution.

Supervised object or scene classification by deep learning networks is an extremely active field of research thanks to the outstanding performances obtained in computer vision. The key advantage of these networks is their ability to learn simultaneously relevant features and a class discriminative classifier. A possible direction will be to use recent end-to-end approaches that accept images as input and directly generates a set of object bounding boxes as outputs 1, 2, 3

Required skills

We are searching for an outstanding and highly motivated candidate with:

Application instructions

The application consists of a motivation letter, CV (with detailed list of courses related to computer science and computer vision), list of publications if applicable, names and contact details of two references, and transcripts of grades from under-graduate and graduate programs. Applications should be submitted before June 30th, via electronic mail to ducottet@univ-st-etienne.fr


  1. Russell Stewart, Mykhaylo Andriluka, Andrew Y. Ng (2016), End-To-End People Detection in Crowded Scenes, CVPR.

  2. Jiang Wang, Yi Yang, Junhua Mao, Zhiheng Huang, Chang Huang, Wei Xu, (2016) CNN-RNN: A Unified Framework for Multi-Label Image Classification, CVPR.

  3. Jason Kuen, Zhenhua Wang, Gang Wang (2016), Recurrent Attentional Networks for Saliency Detection, CVPR.

Dans cette rubrique

(c) GdR 720 ISIS - CNRS - 2011-2015.