Bilgin's Blog
Reports

 

PDFEffects of Aging over Facial Feature Analysis and Face
Recognition

Report, September 2009

 

PDFFace Recognition Across Ages
Report, April 2010

 

PDFOn Age Estimation by Using Still Face Images
Light Survey + Report, March 2011

 

PDFAge Estimation from Still Images
Report, April 2011

 

My PhD Journal

After a long period of separation, in 2009, I've decided to return to my PhD studies. And after completing the last mandatory credits and the notorious qualification exam, I'm in the PhD thesis stage.

While I was in the struggle of the qualification exam, I thought that it was the bottleneck of the PhD process. Though the exam was ridiculously hard, now I discover that I'm in the uncharted territory. Finding a meaningful PhD thesis subject and working on it is extremely hard.

By this journal, I'll try to keep things in one place. Besides, the material I'll collect here, may be some helping hand to other researchers.

My Research Field

Age Estimation from Still Face Images

 Aging

KEYWORDS : age estimation, face recognition, machine learning, pattern recognition, image processing, human aging process, age progression, biometrics

Why do I Make Research on Age Estimation?
When I re-started my PhD studies, I had the ambition for astronomy and space sciences. The plan was working on Pattern Recognition techniques applied on astronomy. After consulting with my professor Bulent Sankur, and also realizing that the field is not a very fertile ground for pattern recognition, I started looking for a new research field.

It was again the idea of Bulent Sankur that human age related studies may be a good idea to start. At first it didn't appeal me very much; not just because of the ironic situation that the "aging studies" would be held by the oldest PhD student in the neighborhood, but mainly because it was not as striking as astronomy.

I slept on the idea for a time. And one day, I watched "Smokey and the Bandit" and "Smokey and the Bandit II" (on the same night) again after 20 years. The second film was shot 4 years after the first one. While watching the films, I observed the eventual aging process of the actors and actresses. Might it be an interesting field?

Smokey and the Bandit
Burt Reynolds and Sally Field from the movie Smokey and the Bandit

At that night, after watching two films together, in my dream I saw Burt Reynolds and Sally Field and some other characters from the movie. I saw their faces changing with passing time, and again becoming younger, and then becoming old again. The last image from my dream was again the youthful face of Sally Field.

In the morning, I decided on two things:
  1. The time passes too fast, so appreciate it genuinely (regardless of how old you are)
  2. I would study on aging
Bülent Sankur
Bulent Sankur

Bulent Sankur is one of the most distinguished professors in Turkey and also one of the leading researchers in Signal / Image Processing field in the world.

He's the founder / director of BUSIM laboratories in Bogazici University and also chairman of Electrical and Electronics Engineering Department of Bogazici University. He has also hundreds of other titles and awards which I don't have any idea of, and thousands of international papers which were cited millions of times.

In addition to these highly successful career credentials, he's also a good reader of books, listener of quality music, follower of fine arts, has gusto for fine food and has a huge knowledge on well .., nearly everything. He can achieve this by his exceptional self-discipline, and he expects this from all his students including me.

Although he has high standards for PhD studies, and sometimes it may be very tiresome to cope with it, I recommend all PhD candidates to work with people like Bulent Sankur.

Click for his personal page.
Age Estimation by Using Still Face Images
Now I'm mainly concentrated on age estimation by using global features. The information that you may be interested is grouped in three sections:

1. Background information on Age Estimation

This page contains some background information on age estimation and a literature overview.

Click here.

2. My approach for age estimation problem

This page contains my approach to the problem. Algorithms, some results, observations.

Click here.

3. Hollywood Database

Now, I'm in the process of constructing a custom database for my researches. It's not yet complete but you can follow its progress.

Click here for Hollywood database.
Milestones
2009, Jan
Starting PhD
2009, Mar
Started the courses EE572 and EE58A
2009, May
Completing courses EE572 and EE58A
2010, Jan
Failed the Qualifying written exam
2010, Mar
Workshop on Biometrics Related to Aging, Canterbury England
2010, May
Succeeded the Qualifying written exam
2010, June
Succeeded the Qualifying oral exam  - with the condition of taking another 2 courses
2010, Oct
Started the courses EE573 and EE577
2011, Jan
Completed the courses EE573 and EE577
   
About  |  Affiliate Program  |  Sitemap  |  FAQ's Privacy Statement  |  Terms Of Use
Copyright 2009 by Bilgin Esme