[maemo-community] GSoC Student Proposal

[From: Sudheendra Bhat]

Hello All,

I am planning on submitting a Student Proposal to the Google Summer of  
Code, and my proposal is pasted below. Any comments/suggestion will be  
appreciated.

Thanks,
Sudheendra Bhat

     * Project title.
           o Text optical recognition on Photo Images
     * Project Abstract:
           o The idea of optical character recognition will be  
extended to recognize the characters available on a Photo Image. The  
main advantage being able to use the character information available  
in the photo image with different services like OpenStreetMap (Say if  
we have the photo image with the street names, then it can be  
recognized and used with OpenStreetmap for tracing the street on the  
map), Language Translators (we can use the character information  
available on the Photo with translate.google.com), in Social Blogging  
applications to provide the history of any specific event (Say, my  
Sunday hangout at Washington DC, all different information available  
on the photos taken during the whole of Sunday can be used to tag the  
events automatically with this service - Context Awarenss) and many  
more related applications.
     * Project Full Discription (Technical):
           o Generally a photo image may consist of character  
informations with meaningful context and this information can be  
helpful for variety of purposes.In this project an attempt will be  
made to capture the character information available on any Photo  
Image. The character infromation available on a Photo can be captured  
by extending the concept of optical character recognition. Like  
traditional OCR concepts, we would be requiring to have some sought of  
Neural Network/Hidden Markovian Model to train our program with set of  
character types. Along with this, we need to also provide some  
techniques to process the image to identify the character regions and  
process (Recognize) those characters from our trained program. Once we  
have the working module for recognizing the character information from  
the Image, then we can enhance the module by adding some service  
related features (i.e. expose the services so that any third party  
application or program can use this feature of character recognition  
on their environment).The module which we develop for this project can  
be easily integrated with Maemo framwork and will have an efficient  
usage on the Internet Tablets or any handheld device platform.
     * Project tasks and deliverables:
           o The project deliverable will be mainly a program to  
recognize the text characters from a Photo Image (at this stage). And  
future enhance can be made to provide services to third party programs  
may be in the form of a Web Service.
           o My tentative plan/tasks for the execution of this project  
will be:
                 + Community Bonding (meeting and discussing the  
technical aspects of this projects with the Maemo team members),  
understanding the Maemo framework with respect to the current  
requirement.
                 + Planning to use Cairo Library (which is supported  
in Maemo), GDK, GDKPixBuf Libraries for the Image processing related  
tasks. So I would need to spend couple of days to get the experties on  
these Library usages.
                 + Photo Image pre-processing algorithm development -  
for parsing the Photo Image ( Either JPG, TIFF or BMP format) and to  
get the necessary basic information. Also to get the image in the  
format in which my recognition program can understand.
                 + Developing the training/Learning algorithm ( I am  
planning on using either Backpropogation Algorithm (Feed forwad  
algorithm) or some Hidden Markovian Model based algorithm for this  
purpose).
                 + Application Development (A Gtk+ based application  
with Hildon support, which uses the trained data sets on the Photo  
Images which comes as a input to the program to recognize the Text  
Characters available on the Image).
                 + Upon completion of these above steps, I will use my  
experience with the Web Services to develop a service oriented  
architecture for the recognition program and expose relevent api's so  
that other 3rd party application can access and use them on their  
applications.
           o As I have the basic ideas on what needs to be done and  
hands on experience with the libraries I will be using in this  
project, I predict that I should be able to complete the standalone  
program of Text Character recognition within the first 2 months  
(approximately giving 35-40 hrs a week) of Summer and remaining time  
of Summer I am planning on working on the Web Service implementation  
for this project.
     * Do you have other commitments during the summer of code coding  
period ? If yes, what kind of commitments ?
           o NO
     * Why are you the right person for this project ?
           o I have some hands on experience with respect to the  
Recognition system, as I had done a project on Optical Character  
Recognition using the Neural Network based algorithms. During my  
internship I did work on the development of context aware application  
on Hildon Framework where I learn about DBUS, Hildon Framework, Bluez  
and most of the libraries used on a Internet Tablets.I also have  
experience progragmming with respect to the mobile platform. Above all  
I have the zeal to contribute towards the opensource community.
     * What are the benefits of your project for the Maemo Community ?
           o This project is targetted to run on a Handheld platform  
as it can give many useful services (as described in the abstract) to  
the user. And Maemo framework already provides the necessary library  
support required for this project, so that this project can be easily  
integrated with the framework with no extra overhead. Also, this  
project gives birth to many new topics of interests like using the  
services from this projects in Social Community like blogs, photo  
gallery, and implementing many new context aware based applications  
targetting the handheld devices.
     * ABout Me:
           o To introduce myself, I am Sudheendra Bhat currently  
pursuing my masters in Computer Science from Dept of Computer Science,  
George Mason University, VA.
             I am interested in this project as I am planning on doing  
my thesis in the areas of Mobile Computing and this project should  
defiantly be a good kick of for the same. Also I have been working on  
mobile platform (Moblin and IPhone) for last one year now and I am  
very much interested in contributing to this field and above all I am  
interested in contributing to the OpenSource community and I think  
this would be a wonderful opportunity for me to do so.

             Here is some more information about my background and  
work experiences,

             I am currently working as Graduate Teaching Assistant in  
Dept of Computer Science, GMU where I am teaching Object Oriented  
Programming with Java to the Undergraduate Students during their Lab  
Sessions.

             I did my Summer Interns 08 at Intel Corporations ,  
Hillsboro OR where I worked on research and development of Context  
Aware Modules for Handheld devices in Moblin Framework (With  
Hildon,DBUS support). My work as an intern was well appreciated with a  
Goodie Drawer Level-2 award for my work during my Internship.

             I have a strong background in the fundamentals of  
Computer Science and extensive programming experience. I have 2 years  
of work experience from Wipro Technologies, Bangalore, where I had  
worked on the design and development of Network Management System and  
Element Management Systems. I have extensively programmed in C, C++  
and Java, J2EE. I can work in both Windows and Unix  environments and  
have a good knowledge of the API’s.

             Also I worked as a Graduate Research Assistant for a year  
with my Professor Dr. Mark Pullen on the development of Joint Battle  
Management Language. This project mainly used Core Java and Java web  
services (Using Axis 1.4) along with mySql as a back end tool. The  
project has been tested and further development is underway. This is  
an OpenSource Project and you can find out more information about JBML  
from, http://netlab.gmu.edu/JBML

             As a part of my course work, last fall 08 I did a project  
called E-Ticket Verification using the Handheld devices. The entire  
project was developed on Moblin framework and I had used Java, and  
Java Webservices for the implementation of the Project. I am working  
on publishing the paper related to my work with the E-Ticket  
Verification project and I will be submitting my paper by end of  
April.I am currently working on 2 main projects as a part of my  
academia work, first one on buidling a Compiler for a Language called  
Tiger, and second project is on Buiding a Gene Prediction algorithm.
           o You can find more info about me at,
             http://c4i.gmu.edu/~sbhat2