DIGI-ACCESS

PROBLEM STATEMENT AND SCOPE

Problem Statement:

To develop a application which will control the access of system using image recognition.

What is the problem it intends to solve?

It is difficult to maintain the physical id or access cards,registers etc. for the identity purpose. Using this app we can digitalised the existing system. It can be customized according to the need and use of the system, for instance, it can be used in a college for facilitating the access to the laboratory,library etc. only to the students of the college or access to the college office only to staff members and teachers.

Why it needs to be solved?

It will simplify the existing access system where the physical cards and entry registers are need to be manually maintained and updated. The access cards can be damaged or manipulated by unfair means and it does not provide 100% authenticity.

Scope Of Project

It includes capturing the image of a person,identify it and providing the access to the authorized user.The database can be updated for making new entry for a new user.

Novelty Of Idea

Using OpenCV for recognising the image of a person and providing the control access.

Relevance And Application

SOLUTION

Technology Layer

Platform

Language

R&D

R&D element in the project will be implementing Computer Vision using OpenCV.
Computer Vision is the transformation of data from a still or video camera into either a decision or a new representation.
OpenCV is a open source computer vision library and was designed for computational efficiency and with a strong focus on real time application.
We are using OpenCV's Eigenface algorithm. It is simple face recognition algorithm and is used in many advanced methods. One reason Eigenface is so important is that the basic principle behind it,PCA(Principle Component Analysis) and distance-based-matching appear over and over in numerious computer vision and machine learning applications.
It is written in optimized C and can take advantage of multicore processors.

2 members like this

Comment

Comment by Prerana Polekar on August 31, 2011 at 11:57am: Yes sir we'll keep updating you...

Comment by Kaushik on August 31, 2011 at 9:10am: Great. Let me know how it progresses.

Comment by Prerana Polekar on August 30, 2011 at 3:29pm: Yes sir we got the face-detection(not recognition) part working on the android phone today only...Its detecting one face at a time and displaying the no of frames the camera is taking per second...This code is entirely in java,so we were thinking if we could get the recognition part also in java we can eliminate the ndk complexities...We found that there is java extraction for OpenCV(JavaCV) which has the java wrapper classes for opencv functions so we can code the entire face recognition part in java and it works with android also...The problem is there is no javadoc like documentation available which can give us information about which all functions javacv has,only samples available for the info.... We'll search on it..

Comment by Kaushik on August 30, 2011 at 7:39am: Any update on the opencv setup. While we may not meet this weekend due to festivals, lets try and make some progress on the projects.

Comment by Prerana Polekar on August 19, 2011 at 8:16pm: Kaushik Sir,

After searching through the net for object recognition using opencv haar training seems to be most reliable and maximum accuracy giving process.We tried to search for its alternative but other methods are mostly shape or colour based( We are able to detect a red ball using the webcam) which will not be much of the use in our prototype.Sir should we start working towards haar training??

I have understood the basic process of haar training but some things are still difficult to get,I am sure as we move forward & read more and more about it, it'll be much more clear..

Comment by Prerana Polekar on August 12, 2011 at 9:30pm: Yes Sir we'll study it more and get the concepts & process more clear.

Comment by Kaushik on August 12, 2011 at 3:01pm: Great, continue with that for the current week so that you are clear on the process. You could then extend to other objects and see how it works.

Comment by Prerana Polekar on August 12, 2011 at 8:48am: Kaushik Sir, I have studied the code for face recognition app and understood the flow of the program.Got better grip over OpenCv.By the end of this week we'll be atleast thorough with the face recognition concept of OpenCv. Sir we are doing the R&D for object recognition, till now haar-training seems to be the only way for it,we'll search more..

Comment by Prerana Polekar on August 5, 2011 at 5:25pm: Yes sir that will do...We will study the code and get the details.

Comment by Kaushik on August 5, 2011 at 4:50pm: Great. Before you start training object recognition, spend a day to understand the code for the recognition of faces and how that is working. The url has more details about the various parts of the app.