ORGANISATION/COMPANYUniversité Clermont Auvergne
RESEARCH FIELDComputer scienceMathematics
RESEARCHER PROFILEFirst Stage Researcher (R1)
APPLICATION DEADLINE28/06/2020 00:00 - Europe/Brussels
LOCATIONFrance › AUBIERE
TYPE OF CONTRACTTemporary
HOURS PER WEEK35 H
OFFER STARTING DATE01/10/2020
IS THE JOB RELATED TO STAFF POSITION WITHIN A RESEARCH INFRASTRUCTURE?Yes
Subject: Automatic objects detection and counting in videos of a production
Supervisor: Chafik Samir
Laboratory: LIMOS UMR 6158
Email and phone: firstname.lastname@example.org
Abstract (up to 10 lines):
Detecting and classifying multiple forms of objects in a video stream is becoming a key step in
different artificial systems. For example, automatic supervision and detection by learning
methods are increasingly used for decision support in industrial, agricultural, food, etc.
production systems. The main challenge consists in finding a method that would be able to
distinguish the classes (categories) of objects while detecting them properly. In this project we
propose to develop a unified framework that combines the two techniques using Bayesian
modeling. This will be made possible witha stochastic process formulation for predicting the
form with prior (learning) as well as its class (prediction). Different datasets are available and
will be used to first learn a model which will build a probability map (with uncertainty) and
then separate the predicted areas of objects.
The candidate should be familiar with standard machine learning methods with a solid
- Programming with python (matlab or R).
- Numerical methods and/or stochastic modeling
Artificial Intelligence, Gaussian processes, Machine learning, Bayesian inference
Description (up to 1 page):
Detecting and classifying multiple forms of objects in a video stream is a key step in different machine
vision systems. The challenge is to propose a method that would be able to distinguish the classes of
objects while detecting them properly. For example, automatic supervision and detection by learning
methods are increasingly used for decision support in industrial, agricultural, food, etc. production
systems. Such a need to automate quality control tasks and to fill the lack of expertise among users raises
very important scientific obstacles. One of the main reasons for this scientific interest is the need to
choose robust models on different criteria capable of providing powerful predictions.
Within the framework of this project, we aim to develop accurate methods that will be able to: (i) extract
relevant patterns (characteristics) from the data, (ii) learn reference shapes, (iii) analyze and compare
any new observation with the reference, (iv) infer on a state, and (v) propose a categorization. All these
steps will be regrouped in a single unified framework based on a stochastic formulation for a
probabilistic decision taking into account uncertainty. Indeed, several recent works have shown that
learning methods, in particular methods based on stochastic optimization and modeling, are the most
suitable when it comes to automatic anomaly detection or variation/growth/changes characterization.
Different datasets will be used to learn and test each model.
References (up to ½ page):
- C. Samir, J-M. Loubes, A.-F. Yao, F.Bachoc. Learning a Gaussian Process Model on the Riemannian
Manifold of Non-decreasing Distribution Functions, PR-ICAI 2019.
- C. Samir, I. Adouani, Regression on Riemannian manifolds, Applied Mathematics and Computation
- O. M. Essid, C. Samir, H. Laga. Automatic Detection and Classification of Manufacturing Defects in
Metal Boxes Using Deep Learning. PLoS ONE , 2018.
- Giovanni Maria Maggioni and Marco Mazzotti. Stochasticity in Primary Nucleation:
Measuring and Modeling Detection Times. Crystal Growth & Design 2017.
How to candidate?
Applications from abroad are welcome. For questions please feel free to contact Chafik Samir (E-Mail:
email@example.com). Applications including a CV, a recommendation letter, a list of publications if
any, and descriptions of research and teaching experiences will be appreciated. The application is
requested in electronic form as a single compressed file.
REQUIRED EDUCATION LEVELOther: Master Degree or equivalent
EURAXESS offer ID: 528439
The responsibility for the jobs published on this website, including the job description, lies entirely with the publishing institutions. The application is handled uniquely by the employer, who is also fully responsible for the recruitment and selection processes.
Please contact firstname.lastname@example.org if you wish to download all jobs in XML.