Major Academic Projects

This section hasn't been updated for over a year. Please see my resume instead.

Image Search
Implemented bag of features based image search and I experimented with SIFT descriptors using kmeans and hierarchial kmeans clustering algorithms for generating code-words and further preparing the term document matrix. Improved the results using term frequency inverse document frequency and query expansion techniques.
Technologies used: OpenCV
Guide: Prof. CV Jawahar

News Video Analysis
Processed low quality news channel video recordings and performed key-frame extraction, logo detection and crawler extraction. Proposed a novel edge detection based approach for news logo segmentation which gave good results on all our test data.
Technologies used: OpenCV, ffmpeg
Guide: Prof. CV Jawahar

Registers for Office Automation:
In traditional offices, clerks use long book registers for recording transactions. Such offices can be easily computerized by replacing these registers with systems which are similar to use but that support automation. This project was an effort in providing such a framework. A user could create a register by specifying the desired columns. Automation could be added by writing scripts which are run at the occurance of events (not implemented). I followed the waterfall model of software development.     Read more...
Technologies used: wxPython, mySQL.
Guide: Prof. Kamal Karlapalem, Mrs. Kirti Garg

Spell Checker in any language: Developed a spell checker that can do spell checking in any language. We exploited pythons Unicode compatibility. The user would provide the necessary dictionary which should be in Unicode format. Also Levistein's edit distance is independent of the alphabets that constitute the language. Hence this algorithm was used to get the correct word for every wrong one.        Source code
Technologies used: Python

Inventory Management for IIIT Systems staff:
We, a team of two, developed a portal that facilitates requests for inventory and enables the systems staff to regulate requests and manage allotted inventory, also keeping track of their repair status. First attempt at system development following the MVC (model, view, controller) paradigm. This was a precursor to the major project on registers.     Source code
Technologies used: web2py, Python