Building something beautiful

- I develop different image recognition techniques for Aurora Solar to do site designs remotely and more effectively.- Responsible for fulfilling the grant - http://energy.gov/eere/sunshot/project-profile-aurora-solar-inc-incubator-10

-Implemented a state of the art Iris Calibration algorithm to accurately find the Iris locations in face images. Achieved anaccuracy of 95% on 2500 face images.-Developed a semi automation framework with detection and tracking(NAC, rigid transformation)for a video annotationtool expediting the process of creating ground truths for videos.

I am working on the project of pedestrian and bicycle detection from surveillance videos.- Did an extensive research on pedestrian detection algorithms of the past 10 years.- Segmented foreground and background in surveillance videos using GMM to avoid processing unimportant frames. - Developed the backend for annotation tool for videos which can segment bikes and pedestrians automatically with minimum user interference.- Detected fast moving objects like birds to avoid possible false positives for pedestrian detection

Was part Image and Video Analytics team in Infosys Labs. Worked on Photo Bill Pay and DEEP projects.Photo Bill Pay - 'Photo Bill Pay' is a application which facilitates users to pay their bills just by clicking a picture of the bill with their mobile. -I developed the backend which is document recognition engine, worked on different OCR's like ABBY, iDRS, Tesseract. -Developed an app i Android for blur detection of the captured document images