My graduation research: Facial Age Estimation

Framer: Imed Bouchrika

Colleague: Mohamed Lamine Tebib

In this project , I have developed a system specifically to approximate age range of person aged between 18 and 90 years. During the development of this system, I found that I should pass by many of the most important basic stages:
As a first stage I used a database containing 180 human face images , and each image has the corresponding age of the person. Then I analyzed each of them to extract the largest possible number of features by using a texture-based method : Local Binary Pattern.
The second phase is the training process, where I have proposed a hierarchical pattern in which as first phase I extracted Special Features for people older than 40 years, as Ill for people younger than 40 years old. Afterwards, I determine the first field, I re-training process so as to get the special features age category more accurate than the first, and so on ….
The last stage is the application’s phase, where I will insert the image of the person that I want to estimate his age, I analyze his face by the same way. All what I should to do in the last step is to find the nearest features that Ire obtained during the training phase, to estimate the age range of this candidate person by using the KNN classification rule.


Languages Used: MATLAB, Java
Database Used: BW_Kennedy
Methods Used: LBP (Local Binary Patterns), KNN Classification.

1 Comment

  1. hakim hassani

    lah ybarek kho

Leave a Comment About This Post :)