Manjunath Narayana

MANJUNATH NARAYANA

990 N Pleasant St, Apt E06, Amherst, MA- 01002

(413) 546-1629 (home) ● (413) 548-0821 (cell) ● ● http://vis-www.cs.umass.edu/~narayana/

PROFILE:I am a machine learning and computer vision researcher seeking research and development positions in the industry. My research interests are in object detection and tracking in surveillance videos, motion segmentation, face detection, and object recognition in images.

EDUCATION:University of Massachusetts, Amherst expected 2013

PhD Candidate, 5th year GPA – 3.83/4.0

University of Kansas, Lawrence, Kansas

M.S. Computer Engineering GPA - 4.0/4.0 July 2007

Awarded Honors for my thesis and research work

Thesis: Automatic segmentation and tracking of moving objects in video

URL: http://vis-www.cs.umass.edu/~narayana/NarayanaMastersThesis.pdf

B.M.S. College Of Engineering, Bangalore, India

B.E Electronics and Communication Top 5% of class July 2004

PROFESSIONAL: Computer Vision Intern, MathWorks, Natick, MA May 2012- Aug 2012

Development of new algorithms for the MATLAB Computer Vision Systems Toolbox (CVST)

Responsible for adding new algorithms into the CVST module of MATLAB
In-charge of feasibility study, requirements collection, design, and development of the new additions

Computer Vision Engineer, Pixblitz Studios, Cupertino, CA May 2008- Dec 2008

Development of a new advertising solution for video broadcasts based on object detection in scenes

Responsible for research and development of a new ad insertion method in video broadcasts in a start-up company of 5 employees
Was part of initial core technology development team of 3 members and in-charge of the image processing subsystem
Developed a proof-of-concept system to demonstrate the new technology
Regularly interacted with customers and VCs to understand the market requirements and shape the company’s business strategy

RESEARCH:Graduate Research Assistant, University of Massachusetts

Segmentation of interesting objects from moving camera videos Jan 2012 – present

Exploratory work on segmentation of interesting objects in videos taken from hand-held cameras
Working on combining information from various sources including estimated camera motion parameters, optical flow observations, and object appearance models to achieve efficient segmentation of objects

Background modeling in video sequences using spatial support from neighboring pixels

Jan 2011 – Dec 2011

Improved the latest background subtraction algorithms by incorporating spatial dependency between pixels and adaptive kernel variance selection for each pixel
Combined the most successful probabilistic models for background modeling with the latest texture features to achieve state-of-the-art results on a standard benchmark dataset
Authored an accepted paper at Computer Vision and Pattern Recognition (CVPR) 2012 conference (acceptance rate – 26%)

Graduate Research Assistant, University of Massachusetts

Automatic classification of underwater plankton images for ocean ecology studies Sep 2007-May 2008

Worked on the difficult but important task of developing algorithms for automatic counting and classification of plankton populations in the ocean from poor quality underwater images
The project involved both research into image processing algorithms and system development in matlab, perl, and html
The project was a coordinated effort by 15 researchers at the University of Massachusetts, Louisiana State University, and Bigelow labs, Maine.

Graduate Research Assistant, Dept of EECS, University of Kansas Jan 2006 – May 2007

Automatic segmentation and tracking of moving objects in video

Master’s Thesis dealing with real-time segmentation of moving objects

Devised a new Bayesian-based tracking algorithm for automatic tracking
Authored a CVPR workshop paper describing the algorithm and its success
Built the foundation for video tracking research at KU, no prior work existed in this area at KU

Graduate Research Assistant, HABS projects May 2005 – Dec 2005

Lead Systems Engineer for the Honeywell Corp. funded High Altitude Balloon System (HABS) project
Led a team of 5 members for design, development, and testing of HABS system
Managed flight logistics of the (new) Third generation HABS system
Doubled the efficiency of power subsystem hardware through better research and design
The lighter, more efficient, extremely robust and reliable system is now in use at KU

PUBLICATIONS: M. Narayana, A. Hanson, E, Learned-Miller, “Background modeling using adaptive pixel-wise kernel variances in a hybrid feature space”, IEEE Computer Vision and Pattern Recognition (CVPR), 2012

M. Narayana, A. Hanson, E, Learned-Miller, “Improvements in Joint Domain-Range

Modeling for Background Subtraction”, British Machine Vision Conference (BMVC), 2012 (in submission)

G. B. Huang, M. Narayana, E. Learned-Miller, “Towards unconstrained face recognition”, IEEE Computer Society Workshop on Percepstual Organization in Computer Vision, 2008.

M. L. Gruman and M. Narayana, “Applications of classifying bidding strategies for the CAT tournament”, AAAI Workshop on Trading Agent Design and Analysis, 2008.

M. Narayana and D. Haverkamp, “A Bayesian algorithm for tracking multiple moving objects in

outdoor surveillance video”, IEEE International Workshop on Object Tracking and Classification in and Beyond Visual Spectrum, CVPR 2007

POSTERS: L Sevilla-Lara, M. Narayana, E. Learned-Miller “Distribution Fields: A Flexible Representation for Low-level Vision Problems, MIT Scene Understanding Symposium, 2011

SKILLS:Programming - C, C++, MATLAB, CUDA, Java, Java RMI, Perl

Operating systems - Windows, Linux

PROJECTS: Face detection and pose estimation

- Trained a boosting (Adaboost) cascade face detector using openCV to study the shortcomings of existing detectors

- Developed models to estimate face pose from images and investigated the effect of pose on face recognition tasks

Multi-tiered server system to simulate an online book store

- Implemented a multi-tiered server system that interacts with clients and accepts book orders.

- Developed using distributed programming principles of Java RMI

Automated text categorization using Support Vector Machines

- Used Support Vector Machines (LIBSVM) to categorize a two-class text corpus

Word sense disambiguation using Naïve-Bayes classification

- Disambiguation of word pairs, implemented using C++

Building a Information Retrieval engine from scratch

- Course project for Information Retrieval class – aimed at learning how search engines work

Feature elimination in breast cancer data with Recursive Feature Elimination method using LIBSVM

- A feasibility study in use of Support Vector Machine (SVM) based RFE feature elimination in breast cancer patient data

Hierarchical Neural Network for classification of Kansas Geological Survey rock facies data

- Used a hierarchical Neural Network approach to classify rock data to identify oil bearing rocks

Analysis of Multiple Sequence Alignment methods

- A detailed study and performance comparison of various Protein Alignment algorithms

Quickly learned theory and implementation of diverse algorithms: Saga, Clustal and HMMer

ACTIVITIES:

Member, International Programs Office staff for International Student Orientation
Treasurer, International Student Ambassador Program, KU
Program Coordinator and committee member of BMS College Mountaineering Club
Captained teams in Boxing, Football, and Hockey at Rashtriya Indian Military College (RIMC)
Mentored junior cadets by planning daily activities, coaching and motivating a section of 50 students in my capacity as Officiating Section Commander at RIMC
Part of the RIMC team which participated in adventure sports (mountaineering, parasailing, whitewater rafting)

ACHIEVEMENTS:

Selected to join the Indian Air Force. Secured rank 9 in the country among all candidates
Gold medalist and College colors in Boxing at RIMC
Secured First rank and was the lone candidate to qualify from Karnataka state to join the Rashtriya Indian Military College (RIMC) through a country-wide selection process

CV - Manjunath Narayana