Jason L. Clemons
Whitmore Lake, MI 48189
734-***-**** US
abo7ax@r.postjobfree.com
OBJECTIVE: Secure a position that will allow me to develop and design computer
architectures for mobile
vision applications.
EDUCATION:
University of Michigan, Ann Arbor, MI
08/08- present Degree: Ph.D. Computer Science and Engineering
Concentration: Computer Architecture
GPA: 8.593/ 9 (4.0/4.0)
Expected Graduation Date: May 2013
09/01-08/02 Degree: M.S. Computer Science and Engineering
Concentration: Computer Architecture
GPA: 4.0/4.0 Graduation Date: August 2002
Michigan Technological University, Houghton, MI
09/96-05/00 Degree: B.S. Electrical Engineering
Concentrations: Computer Hardware, Control Systems
GPA: 4.0/4.0 Graduation Date: May 2000
RELEVANT COURSES:
Computer Architecture Parallel Computer Architecture Microarchitecture
Computer Vision (Visit) Advanced Computer Vision (Visit) Algorithms for Robotics (Visit)
Machine Learning (Visit) Intro. to Artificial Intelligence Principles of Real Time
Computing
RESEARCH & PROJECTS:
EFFEX: A Processor for Mobile Computer Vision
Ongoing research. University of Michigan, September 2009 to Present.
Advisors: Prof. Silvio Savarese & Prof. Todd Austin
This research focuses on developing computer architectures for mobile vision
applications. The objective is to
provide 1000x improvement in mobile vision computation. This will be made up of 10x
improvement in energy
efficiency and 100x in performance. We are developing a heterogeneous multi-core
architecture that supports
custom functional units and memory systems optimized for the mobile computer vision
space.
Visual Sonification for the Visually Impaired
Ongoing research. University of Michigan, September 2008 to Present.
Advisors: Prof. Silvio Savarese & Prof. Todd Austin
This research focuses on developing computer vision algorithms, sensor fusion algorithms,
3d audio algorithms, and
computer architecture for an embedded system that allows the blind to localize, navigate
enclosed areas and
recognize specific objects using audio cues. I am responsible for vision algorithms,
computer architecture,
embedded software and system integration. I also oversee the work of undergraduate and
other graduate research
for the development of the sensing and 3d audio systems.
Identifying Occlusion Edge Pixels Using Disparity and Low Level Image Cues
Preliminary Examination Research, University of Michigan, September 2009.
Advisors: Prof. Silvio Savarese & Prof. Todd Austin
This research used noisy stereo images from a low cost/quality camera to identify the
occlusion edges from a scene.
The novel component focused on using disparity maps from Gaussian Scale Space as cues to
occlusion edge
location. I developed the algorithm for finding the Gaussian Scale Space Disparity Metric
(GSSDM) that is used for
with edge detection techniques such as gPb to classify occlusion edges.
SIFT performance evaluation on GPGPU using CUDA
Parallel Computer Architecture Class Project.
University of Michigan, Fall 2008.
Team member: Ahsen Tahir Advisor: Prof. Todd Austin
This research focused on analyzing the performance speedup of Scale Invariant Feature
Transform (SIFT) when
implemented using Compute Unified Device Architecture (CUDA). This project implemented
the first 2 phases of
David Lowe s SIFT and investigate the performance speedup and bottlenecks in the
implementation. I was in
charge of SIFT implementation on in C and in CUDA. We were equally responsible for the
analysis.
PUBLICATIONS & POSTERS:
MVSS: Michigan Visual Sonification System
IEEE Emerging Signal Processing Applications (To appear), 2012
Authors: Jason Clemons, Sid Yingze Bao, Vinay Sharma, Silvio Savarese & Todd Austin
Visual Sonification is the process of converting visual properties of objects into sound
signals. This paper describes
the Michigan Visual Sonification System (MVSS) that utilizes this process to assist the
visually impaired in
distinguishing different objects in their surroundings. MVSS uses depth information to
first segment and localize
salient objects and then represents object appearance using histograms of visual
features. The histogram of a
segmented object is then converted to a sound signal, the volume and 3D placement of
which is determined by the
relative position of the object with respect to the user. The system then relies on the
considerable discriminating
power of the human brain to localize and ``classify'' the sound, thus enabling the user
to distinguish between visually
distinct object classes.
Visual Sonification
Poster, Gigascale Systems Research Center Projects Review, November 2011. Best
Poster/Demo Award Winner
Authors: Jason Clemons, Sid Yingze Bao, Mohit Bagra, Max Seiden, Erik Rainey Vinay
Sharma, Silvio Savarese &
Todd Austin
Visual Sonification is the process of converting visual properties of objects into sound
signals. The poster described
the visual sonification process and was accompanied by an early prototype of the Michigan
Visual Sonification
System (MVSS) which utilizes TI s vision framework running on a TI Blaze 4430. The system
did not fully utilize
the capabilities of the TI Blaze but was still able to demonstrate the difference in
sonification results when the
system is applied to various objects. In this demonstration we used a sock monkey and a
toy car that allowed users
to hear the difference in the generated audio signatures. MVSS uses depth information to
first segment and localize
salient objects and then represents object appearance using histograms of visual
features. The histogram of a
segmented object is then converted to a sound signal, the volume and 3D placement of
which is determined by the
relative position of the object with respect to the user. The system then relies on the
considerable discriminating
power of the human brain to localize and ``classify'' the sound, thus enabling the user
to distinguish between visually
distinct object classes.
MEVBench: A Mobile Computer Vision Benchmarking Suite
IEEE International Symposium for Workload Characterization, November 2011
Authors: Jason Clemons, Haishan Zhu, Silvio Savarese & Todd Austin
We put together a set of benchmarks, some single threaded and some multithreaded, to
evaluate embedded processor
performance in commonly performed mobile vision applications. Using this suite we
examined the performance of
various processors architectures including Intel Atom and Arm A9 while the input size and
thread count was varied.
We also developed a new metric for measuring the regularity of branching to determine the
affinity to GPGPU
architectures. Finally we proposed architectural ideas that could be used to increase
performance. Website:
http://www.eecs.umich.edu/MEVBench
EFFEX: An Embedded Processor for Computer Vision Based Feature Extraction
Design Automation Conference, June 2011
Authors: Jason Clemons, Andrew Jones, Robert Perricone, Silvio Savarese & Todd Austin
This work focused on the development of an embedded processor for computer vision based
feature extraction. We
examined three feature extraction algorithms from computer vision and developed a multi-
core embedded processor
architecture to efficiently perform feature extraction. Through custom functional units
and a specialized memory
architecture we achieve a higher normalized performance than current embedded solutions.
Classifying Occlusion Edge Pixels
Poster, Gigascale Systems Research Center Projects Review, September 2010.
Advisors: Prof. Silvio Savarese & Prof. Todd Austin
This research combines the GSSDM research from my preliminary exam with color,
brightness, and texture cues
using machine learning techniques to better classify occlusion edge pixels. Using my
GSSDM algorithm with the
other cues in a feature vector allowed for improved occlusion edge pixel classification.
EMPLOYMENT HISTORY:
09/09 University of Michigan, Ann Arbor, MI
to present Position: Graduate Student Research Assistant
Research Interest: Embedded processor design for computer vision applications
Current Research Projects:
MVSS: Embedded sonification system for people who are blind
EFFEX: Embedded processor for computer vision applications
MEVBench: A benchmarking suite for mobile computer vision
06/11Texas Instruments, Dallas, TX
to
09/11 Position: SummerIntern
Development of visual sonification system for TI development platform
Responsibilities:
-Develop PC application containing full sonification pipeline
-Develop Android sonification application for TI Blaze platform
-Benchmark sonification application
06/10NVIDIA Corporation, Santa Clara, CA
to 09/10 Position: Summer Intern
Development of computer vision algorithms for prototype
Responsibilities:
-Develop robust computer vision algorithms for prototype
-Develop data set for testing of algorithms
-Software development for proof of concept
-Assessment of feasibility
07/00Whirlpool Corporation, St. Joseph, MI
to 12/08 Position:
Embedded Systems Senior EngineerPart-time, 08/0812/08
Position:
Embedded Systems Senior Engineer, 10/06 to 08/08
Development of Controls for Production Appliances
Responsibilities:
-Development of microcontroller evaluation process
-Software Lead for development of sensor-less vector motor control
-Development of software architecture for motor control
-Management of intern for delivery of drivers for motor control platform
Position:
Embedded Systems Project Engineer, 05/0410/06
Development of Controls for Production Washer
Responsibilities:
-Development of BPM motor control software for washing machine
- Evaluation and selection of microcontrollers for washing machine controls
-Roadmap development for future electronic technologies
-Manage interns for delivery of software and hardware projects
Position:
Hardware Engineer- Networked Appliances Group, 05/0305/04
Research and development of embedded solutions for connectivity
Responsibilities:
-Investigate various networking technologies for use in appliances
-Design network interface unit for Polara oven
-Lead of embedded development
Position: Whirlpool Technical Excellence Program Rotational Program, 07/00 05/03
Rotation 4/4: WTEP Whirlpool Sweden AB, Microwave Oven System Engineer, 11/02 05/03
Designing microwave control systems
Responsibilities:
-Software and hardware design for microwave oven control system
Rotation 3/4: WTEP Advanced Electronics, Embedded Controls Engineer, 05/02 11/02
Building tools for embedded controls development
Responsibilities:
-Develop basic Simulink blocks for microcontroller functional units
-Develop base for auto-code generation using Simulink
Rotation 2/4:
WTEP Product Development, Electronics Applications Engineer, 03/0108/01
Cost and quality improvements on control boards for Calypso Washer.
Responsibilities:
-Investigate and resolve quality issue with buzzer
-Investigate transformer quality and suppliers
Design and create real time data acquisitions application.
Responsibilities:
-Learn Visual C++ and implement a data acquisitions system
Rotation 1/4: WTEP Integrated Home Solutions, Software Engineer, 9/00 3/01
Web application development using Java
Responsibilities:
-Develop Java servlet
-Investigating voice control integration
-Develop laundry assistance application
09/99Michigan Technological University, Houghton, MI
to 05/00Research Assistant - Assisted in Auger Observatory Project custom ASIC design
Responsibilities:
-Particle detector circuit design and simulation with Cadence software
06/99Ford Motor Company, Dearborn, MI
to
08/99SummerIntern - Web development using MS FrontPage, PERL, and Netscape
Enterprise Server
Responsibilities:
-Requirements gathering from those who used old application
-Teamwork with another developer to coordinate efforts
-Department web server administrator
09/98Michigan Technological University, Houghton, MI
to 05/00 Learning Center Coach - Aided students in electrical engineering coursework
06/98Delphi Automotive Systems, Flint, MI
to
08/98SummerIntern - In spark plug development researching spark patterns
Responsibilities:
-Investigate and explain sparking pattern
-Contact software vendors to help develop model of spark pattern
-Self directed study of spark patterns
SKILLS:
Electrical: Analog design, Digital design
Software: MS Office, MS Visual C++, Matlab, OpenCV, Android, Ubuntu
Programming Languages: C, C++, Java
RTL Design: Verilog
Microcontrollers/Processors: ST: STM32, TI: OMAP 4430(Arm A9), OMAP 3350 (Arm A8)
Professional Training: Whirlpool Six Sigma, Whirlpool Talent Assessment
ORGANIZATIONS:
National Society of Black Engineers, 2000 to present
Society of Women Engineers, 2001 to present
-Section Web Site Coordinator, 2003 to 2005
-SW Michigan Section Charter Member, 2004
IEEE, 1998 to present
Eta Kappa Nu (HKN), 1999 to present
HONORS/AWARDS:
Margarida Jacome Best Poster/Demo Award - Visual Sonification, Gigascale Systems
Research
Center (GSRC) Annual Symposium, GSRC, November 2011
Scholar Power Award for Scholarship, University of Michigan, 2011 & 2002
Rackham Merit Fellowship, University of Michigan, 2008
nd
2 Degree Black Belt, Sidekicks Taekwondo, 2007
Michigan Governing Boards Award, State of Michigan, 2000
National Science Foundation (NSF) Graduate Fellowship, NSF, 2000
ACTIVITIES:
Whirlpool Employees for Retention and Development, 2004 2008
-Chair, 2004 2005
Botball mentor, Lakeshore High School, Stevensville, MI, 2001 2002, 2004-2008
-Regional Champions - 2007Whirlpool Recruiting Teams North Carolina A&T, University of Michigan, NSBE, SWE, 2001
2008
-NSBE National Conference Booth Lead, 2008
-SWE National Conference Booth Lead, 2007
-Innovating for Market Leadership Seminar SWE National Conference, Presenter, 2004
NSBE Jr. Mentor, Benton Harbor Tech Center, 2002
References Available Upon Request