The recognition and prediction of people behaviour from videos are major concerns in the field of computer vision. A specific class of behavior analysis concerning facial expression recognition attracts lot of attention from researchers and industry in various fields.
State of the art solutions work fine in controlled environments were expressions are exaggerated and the head of the subject stay still, but as soon as the subject moves freely and expressions are natural, the performances of existing systems drop in an important manner. This observation is confirmed by performance evaluation conducted on new datasets (such as RECOLA, GEMEP) where the acquisitions conditions are similar to natural interaction settings.
We look for a PHD candidate in order to study and propose algorithms that analyze human behavior from video in unconstrained environments.
== Required expertise
Strong preference will be given to candidates with experience in Computer Vision and Pattern Recognition and a good knowledge of written and oral English. Background in motion analysis field would be appreciated.
Applicants are expected to have a strong background in Computer Science. Strong programming skills (C or C++) are a plus. French language skills are not required, English is mandatory.
The thesis shall start October 1st in Lille.
They must contain a statement of interest, a CV, a list of publications, if any, and the names of two references.