Computer Science Software Engineer
Resume:
RIZWAN CHAUDHRY
Johns Hopkins University e-mail: ********@***.***.***
Center for Imaging Science Phone: +1-443-***-****
*** ***** **** ***.***.jhu.edu/ rizwanch
Baltimore, MD - 21218
EDUCATION
Johns Hopkins University, Baltimore, MD 2006 - present
Doctor of Philosophy, Computer Science
Johns Hopkins University, Baltimore, MD 2006 - 2009
Master of Science in Engineering, Computer Science
Lahore University of Management Sciences (LUMS), Lahore, Pakistan 2001 - 2005
Bachelor of Science (Honors), Double Major in Computer Science and Mathematics (GPA: 3.91/4.00)
RESEARCH INTERESTS
Classi cation of dynamic visual processes
Simultaneous tracking and recognition of humans in videos
Video tagging and retrieval
Large-scale non-linear data modeling
RECENT RESEARCH EXPERIENCE
Johns Hopkins University September 2006 - present
Research Assistant, Center for Imaging Science
Simultaneous Tracking and Recognition of Dynamic Visual Phenomena Proposed a novel
optimization scheme that jointly minimizes an objective function over the location of a dynamic tem-
plate and its internal state. The method provides excellent tracking, recognition and synthesis results
for human actions and dynamic textures.
Statistics on the manifold of dynamical systems Proposed a theoretically sound as well as
computationally feasible approach for computing means on the space of Linear Dynamical Systems
(LDS) by considering the group action of the space of orthogonal matrices on the space of LDS.
Simultaneous context dependent recognition of objects and actions in television shows
Developed a method for recognizing manipulation tasks and tools from TV shows such as the PBS
Sprouts Crafts shows. A joint model for object and action co-occurrence is learnt from domain videos,
domain transcripts as well as web-based image and language sources. It is then applied to novel test
videos in the same domain to get superior object and action recognition performance.
Semantic searching of very large datasets Proposed a new approximate nearest neighbor (ANN)
method that generalizes Spectral Hashing for non-Euclidean data. The method only requires knowledge
of the Riemannian structure of the manifold or the de nition of a kernel on the space, a signi cant
advantage over Locality Sensitive Hashing and other state-of-the-art methods. The algorithm was
applied to search for activities in a large data-set.
Human Activity Recognition Proposed scale and direction invariant Histograms of Oriented
Optical Flow (HOOF) features to represent action pro les at a particular time instant. The non-
Euclidean time-series thus generated from the activity video was modeled as a Non-Linear Dynamical
System (NLDS). Developed an algebraic method for classi cation of non-Euclidean time-series data
based on the Binet-Cauchy kernels for NLDS using Mercer kernels de ned on the non-Euclidean space.
R. Chaudhry
p. 2
PROFESSIONAL EXPERIENCE
Primary Student Researcher, JHU CLSP Speech and Vision Summer Workshop June 2010 - July 2010
Worked as part of a team of researchers from academia and industry on simultaneous object and action
recognition in unconstrained TV shows.
Research Intern, Mitsubishi Electric Research Laboratories June 2009 - September 2009
Developed algorithms for very fast semantic searching of large scale video datasets.
Software Engineer, Techlogix, Inc., Pakistan June 2005 - July 2006
SELECTED PUBLICATIONS
R. Chaudhry, G. Hager and R. Vidal, Dynamic Template Tracking and Recognition, International
Journal of Computer Vision (IJCV). (Under review).
R. Chaudhry and Y. Ivanov, Fast Approximate Nearest Neighbor Methods for Example-based Video
Search, Video Analytics for Business Intelligence, Springer-Verlag 2012.
A. Ravichandran, R. Chaudhry and R. Vidal, Categorizing Dynamic Textures using a Bag of Dynam-
ical Systems, IEEE Transactions on Pattern Analysis and Machine Intelligence. (To appear).
B. Afsari, R. Chaudhry, R. Vidal and A. Ravichandran, Group Action Induced Distances for Aver-
aging and Clustering Linear Dynamical Systems with Applications to the Analysis of Dynamic Visual
Scenes, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2012.
B. Sapp, R. Chaudhry, X. Yu, G. Singh, I. Perera, F. Ferraro, E. Tzoukermann, J. Kosecka, J. Neu-
mann, Recognizing Manipulation Actions in Arts and Crafts Shows using Domain-Speci c Visual
and Textual Cues, International Workshop on Video Event Categorization, Tagging and Retrieval for
Real-World Applications (VECTaR2011), in conjunction with the IEEE International Conference on
Computer Vision (ICCV), 2011.
R. Chaudhry and Y. Ivanov, Fast Approximate Nearest Neighbors methods for Non-Euclidean Man-
ifolds with Applications to Human Activity Analysis in Videos, European Conference on Computer
Vision (ECCV), 2010.
R. Chaudhry, A. Ravichandran, G. Hager and R. Vidal, Histograms of Oriented Optical Flow and
Binet-Cauchy Kernels on Nonlinear Dynamical Systems for the Recognition of Human Actions, IEEE
Conference on Computer Vision and Pattern Recognition (CVPR), 2009.
HONORS
Reviewer for several top computer vision journals including TPAMI, IJCV, SMC-B and MVA.
Recipient of full tuition fellowship and research assistantship by the Department of Computer Science,
Johns Hopkins University.
COMPUTING SKILLS
Operating Systems: Linux, Windows.
Programming Languages: C/C++, MATLAB, Java
Systems and Tools: Oracle, MySQL, openCV.
RELEVANT COURSEWORK
Computer Vision Random Signal Analysis
Advanced Topics in Computer Vision Linear Dynamical Systems
Machine Learning Statistical Theory
Network Embedded Systems and Sensor Networks Distributed Systems
REFERENCES
Dr. Rene Vidal, Associate Professor, Biomedical Engineering Department, Johns Hopkins University.
Dr. Gregory Hager, Professor, Computer Science Department, Johns Hopkins University.
Dr. Yuri Ivanov, former Senior Principal Research Sta, Mitsubishi Electric Research Laboratories,
currently at Heartland Robotics.
Contact this candidate