Learn all about FACE DETECTION CODE. A Practical Guide to Face Detection using Gabor Feature Extraction and Neural Networks

From the desk of Omid Sakhi Today's Date: Thursday, August 28, 2010

Dear My Friend,
Are you struggling to understand the practical tricks and concepts behind object detection ? Do you know how to start your program?

And How would you fancy doing this WITHOUT having to commit your precious time and efforts developing your program by trial and error?

My name is Omid Sakhi and I am the author of the program Face Detection System for MATLAB. In 2006, I gladly signed up to work on a project for my image processing course at the university. I was studying in my second semester for my master’s degree. The objective of the project was to work on face detection in MATLAB and to present the concept and methods about how it is possible and how to do it properly.

Back then I knew other programming languages like C and C++, but I still had difficulties with MATLAB programming language. Literally, Face Detection System was my first serious attempt to learn MATLAB, face detection, feature extraction and machine learning.

I got prepared to start writing the first line of the program and the moment I opened MATLAB, all my enthusiasm dried up like Sahara   desert. I realised that I don’t know anything about the subject. Needless to say, there weren’t any other code or library for face detection. To cut the story short, it took me three months to research the popular methods for detecting faces and to chose neural networks because of my interests in artificial brain.

Then I uploaded the program in MATLAB Central and to this date I have got  more than several hundred questions in my email about the code and the way it works.

Thus, I decided once and for all to write a complete practical guide for it. This guide that you can get, is the answer to the WHYs and HOWs that I have got so far. It doesn’t go so deep into boring mathematical details. Instead it is practical and starts by describing different parts of the program using the code and the concepts behind it.

Omid Sakhi (MATLAB's Rank : 36 )

Anyway, Is it right for you?

1. You have basic familiarity with MATLAB and at least you have heard of machine learning or neural networks.
2. You want to learn about object detection (object, sign, face, ... )

What is in it for you ?

1. This guide is a practical guide. It means you can barely find mathematical formula in it. Instead you will find lots of concepts, MATLAB commands.
2. It teaches you how to think and write you first lines.
3. It teaches all the tips and tricks that each one may take one week of your time, if you don't know how to handle it.
4. It teaches you how to generate train sets and how to train your neural network. Something that can rarely be found when you download a program.

Table of Contents

• Introduction
• Introduction to MATLAB
• Artificial Neural Networks
  1. A Perceptron Neuron
  2. A Single Layer Perceptron
  3. Multi-Layer Feed-Forward Backpropagation Neural Network
     • Creating It: The Easy Way
     • Input and Output Explained
     • Initializing It
     • Creating It: The Hard Way
     • Training It
     • Testing and Simulation
     • Computing the Error
     • Generating Training Sets
• Feature Extraction
  1. Gray-Level Features
  2. Histogram Equalization
  3. Illumination Effect
     • Background Illumination Effect
     • Adaptive Histogram Equalization
  4. Normalization
• Gabor Feature Extraction
  1. Why Gabor Features?
  2. How to Perform Gabor Filtering
  3. Fast Gabor Filtering
  4. Im2Vec.m
• Face Detection System
  1. Pre-Selection
  2. Search Algorithm
  3. Post-Processing
• Future Work
  1. on Pre-Selection
  2. on Fast Convolutional Neural Networks
  3. on Feature Selection
  4. on Classification