Contribution to Face Analysis from RGB Images and Depth Maps

Résumé

Automatic human face analysis refers to the processing of facial images by machines in order to infer useful information, such as identity, gender, ethnicity, mood, etc. Face analysis has many interesting applications in security, human computer interaction, social media analysis, etc. Therefore, though face analysis is a well-established computer vision problem, it is still an active research topic attracting considerable attention from researchers. The research community mainly aims to develop more robust systems with the ability to fulfill the requirements of current applications. This thesis contributes to a number of face analysis tasks: face verification and identification, gender recognition, ethnicity recognition and kinship verification. Faces from three different imaging supports i.e. RGB images, depth maps and videos are used throughout the thesis. We present novel approaches and in-depth studies for solving and improving the face analysis problem. First, we tackle face verification problem from RGB images. The local binary patterns based face verification scheme has been revised through proposing novel efficient representations, which cope with the original approach drawbacks while improving the verification performance. Next, the problems of identity, gender and ethnicity recognition are investigated from both RGB and depth images. The aim is to assess the usefulness low-quality depth images, acquired with Microsoft Kinect low-cost sensor, in coping with facial analysis tasks. The performance of RGB images and depth maps are compared to show the ability of the latter ones to deal with sever environment illumination circumstances. Furthermore, the thesis contributes to the problem of kinship verification from videos, where the family relationship between two persons is checked by comparing their facial attributes. The dynamics of faces are efficiently coded by the means of spatio-temporal descriptors and deep features. The value of using videos in kinship problem is shown by comparing their performance against that of still images. Throughout the thesis, various benchmark databases are used and extensive experiments are carried out to validate our proposed approaches and developed methods. Besides, the results of the proposed approaches are compared against the state of the art, highlighting our contributions and showing improvements. Future directions for the presented contributions are outlined at end of the thesis.

Citer ce document

Accès au document

Ce document est hébergé sur son dépôt institutionnel d'origine.

Auteur(s)

Statistiques

Consultations : 4

Téléchargements : 0