Exploring the Magnitude of Human Sexual Dimorphism in 3D Face Gender Classification

Xia, Baiqiang; Ben Amor, Boulbaba; Daoudi, Mohamed

Document type :

Communication dans un congrès avec actes

Title :

Exploring the Magnitude of Human Sexual Dimorphism in 3D Face Gender Classification

Author(s) :

Xia, Baiqiang [Auteur]
FOX MIIRE [LIFL]
Ben Amor, Boulbaba [Auteur]
FOX MIIRE [LIFL]
Daoudi, Mohamed [Auteur]
FOX MIIRE [LIFL]

Conference title :

International Workshop on Soft Biometrics, in conjuction with ECCV'14

City :

Zurich

Country :

Suisse

Start date of the conference :

2014-09-07

English keyword(s) :

3D face
Gender Classification
Sexual Dimorphism
Ran-dom Forest

HAL domain(s) :

Informatique [cs]/Vision par ordinateur et reconnaissance de formes [cs.CV]

English abstract : [en]

Human faces demonstrate clear Sexual Dimorphism (SD) for recognizing the gender. Different faces, even of the same gender, convey different magnitude of sexual dimorphism. However, in gender classifica-tion, gender has ...
Show more >Human faces demonstrate clear Sexual Dimorphism (SD) for recognizing the gender. Different faces, even of the same gender, convey different magnitude of sexual dimorphism. However, in gender classifica-tion, gender has been interpreted discretely as either male or female. The exact magnitude of the sexual dimorphism in each gender is ignored. In this paper, we propose to evaluate the SD magnitude, using the ratio of votes from the Random Forest algorithm performed on 3D geometric features related to the face morphology. Then, faces are separated into a Low-SD group and a High-SD group. In the Intra-group experiments, when the training is performed with scans of similar SD magnitude than the testing scan, the classification accuracy improves. In Inter-group ex-periments, the scans with low magnitude of SD demonstrate higher gen-der discrimination power than the ones with high SD magnitude. With a decision-level fusion method, our method achieves 97.46% gender clas-sification rate on the 466 earliest 3D scans of FRGCv2 (mainly neutral), and 97.18% on the whole FRGCv2 dataset (with expressions).Show less >

Language :

Anglais

Peer reviewed article :

Oui

Audience :

Internationale

Popular science :

Non

ANR Project :

Reconnaissance automatique d'attributs faciaux (expression, âge, sexe, ethnie) basés sur les visages 3D

Collections :