About

Hello,

My name is Ali Tarhini and i am a Dev. Lead at a local software company in Lebanon. I am currently working in the domain of HealthCare solutions and building software systems for hospitals and other government institutions in the MENA region. Before healthcare, I used to work on Travel Management Systems and even before that I was a developer for Content Management Systems for building dynamic web content.

I have a bachelor’s degree in Computer Engineering & Informatics from Beirut Arab University and have a Master’s degree in Electrical & Computer Engineering from  the American University of Beirut. My undergraduate thesis was about pattern recognition, specifically about face detection and recognition where I developed, along with my collegeaus, a software package that is capable of finding and detecting human faces in still images and apply two recognition algorithms in order to identify the names behind the faces. My masters thesis was on Automatic Loop Parallelization where I invented a new approach for automatically finding an optimal schedule for running multiple iterations of loops in parallel in order to optimize execution speed and reduce memory overhead.

I am a software geek, so my posts will mainly focus on programming, particularly on the .Net Framework and the web, SQL, and a strong focus on algorithms.

14 Responses to About

  1. sat3o says:
    December 14, 2010 at 2:37 am

    Hello. My name is satryo. I’m a bachelor degree student in Computer Science UGM, Indonesia. Can i’ve ur email contact? I want make implementation of face detection using matlab from viola and jones paper “robust real time object detection” for my assignment, but i could’nt understand some part. Like how to construct weak classifier from rectangle feature? I’m not sure with parity,unknown threshold,and what features to make that equation (hj). Could you give me explanation or sample implementation about viola’s method? Great thanks.

    Reply
  2. Bambang Wahyudi says:
    December 7, 2011 at 11:39 am

    Hi..Ali..nice to know you…
    My name is Bambang, a student of computer science IPB, Indonesia. Currently I’m building a vision based system which can detect dual-tires utilisation on truck-type vehicles. I plan to use haar-like feature extraction combined with SVM classifier. But I still do not fully understand about the haar-like feature extraction … I beg you to describe the technique in algorithmic fashioned (pseudo code) … it would be more helpful if you send me the source code (if you have it) of the technique … my e-mail : mbengmerzy@yahoo.co.id …
    Great thanks.

    Reply
  3. cscturkey says:
    April 6, 2012 at 11:24 pm

    Dear Tarhini;

    Thanks for your good works.I want you to send me Chi-merge discritization program exe. (for IRIS data set) .if ypu can send , I will be glad.

    best regards

    Turkey.

    Reply
  4. Fadol Al Farran says:
    June 16, 2012 at 8:30 pm

    Mar7aba Ali,
    you are working with face detection algorithm? I am a Student in Germany from Lebanon.
    Thanks
    Fadol

    Reply
  5. Fadol Al Farran says:
    June 16, 2012 at 8:46 pm

    I am working now with the Methode Modified census Transform from Ernst and Fröba. They work with 3 stage cascade and winnov update rule.you have a adaboost cascade algorthm for Matlab and winnov update rule?

    Reply
  6. Fadol Al Farran says:
    June 20, 2012 at 9:29 am

    mar7aba Ali,
    i have a question please.
    adaboost carry the histograms of the training images face and non-face with the Pixels of the given image or also with the histogram?
    Thanks very much
    Fadol

    Reply
  7. Lakshmi Srinivas says:
    April 6, 2013 at 3:08 pm

    Hello sir,
    I’m Lakshmi, from India. I’m working on face detection algorithm. I need to know more about the training of faces using adaboost to make the detection of faces more accurate.
    Kindly help,
    Thank you
    Lakshmi

    Reply
  8. essam002002 says:
    June 26, 2013 at 12:04 am

    Hi Ali,
    Regarding your chi merge discretization, why did you set E = 0.5 if its value is less than 0.5?
    Another question: when the frequency of one class equals zero in both bins how do you apply the chi square test in this case? EX: suppose you have two bins b1, b2 and two classes c1, c2
    c1 c2 Total
    b1 0 5 5
    b2 0 6 6
    tot 0 11 11

    Reply
  9. awrad hadi says:
    May 11, 2014 at 7:42 pm

    Hi Mr. Ali
    my name Awrad am preparing for master thesis in discretization algorithms and use one of them in one of inductive learning algorithms I need some complete solved examples in chimerge and chi2 and mdlp and what the different between chimerge and chi2
    this is my email:
    awradhadi@yahoo.com
    please if you could help me in this
    regards

    Reply
  10. Rasheed Sarky says:
    December 29, 2016 at 5:59 pm

    Hello Mr. Ali
    I want a permission to use one of the image that I saw in your work. The picture is about variation in illumination, So may I use it in my thesis ?

    Reply

Leave a reply to cscturkey Cancel reply