CUNY Imaging and Visualization Conference, May 11, 2005

 

Imaging & Visualization Mini-Conference
Bridging the Gap between University Research and Business Applications


Wednesday, May 11th, 2005

Room C198, CUNY Graduate Center

5th Ave and 34th St, New York City

Panel Discussions

Moderator
Michael Grossberg

 

    Panelists       

Ying-Li Tian (IBM)
Michael Chan (GE)
Robin Bargar (City Tech)

Dr. Anurag Mittal (Siemens)

Jeff Young (Leica)

Diego Socolinsky (Equinox)

IBM Smart Video Surveillance System
Dr.Yingli Tian,  Member of Research Staff, IBM TJ Watson Research Center

Abstract:
The increasing need for sophisticated surveillance systems and the move to digital surveillance infrastructure has transformed surveillance into a large scale data analysis and management challenge. Smart surveillance systems use automatic image understanding techniques to extract information from the surveillance data. While the majority of the research and commercial systems have focused on the information extraction aspect of the challenge, very few systems have explored the use of extracted information in the search, retrieval, data management and investigation context. The IBM smart surveillance system is one of the few advanced surveillance systems which provides not only the capability to automatically monitor a scene but also the capability to manage the surveillance data, perform event based retrieval, receive real time event alerts thru standard web infrastructure and extract long term statistical patterns of activity.
Recognizing Temporal Events in Video
Dr. Michael T. Chan, GE Global Research

Abstract:
We describe some ongoing work on recognition of temporal events in video that has surveillance applications. We use a probabilistic framework based on Hidden Markov Models (HMMs) to represent the spatio-temporal relations between interacting objects in a scene, and uncertainty in visual observations, where the observables are semantic spatial primitives encoded based on prior knowledge about the specific event types of interest. Abstracting continuous trajectories into semantic relations enables better generalization to other scenes for which little training data may be available. We demonstrate the effectiveness of the approach using aerial video data and simulated data. At the end of the talk, I will also highlight some technically related works that have other applications in security and human-computer interaction.

Biographical Sketch:
Dr. Chan received his Bachelor degree in Electrical Engineering from Oxford University, and his M.S. and Ph.D. degrees in Computer and Information Science from the University of Pennsylvania. He is currently at GE Global Research, where he has played different roles including computer scientist and project leader. Prior to GE, he was a research scientist at Rockwell Scientific Company and a research associate at USC. His research interests include probabilistic models for visual event recognition, biometric fusion, audio-visual speech recognition, visual tracking,  human-computer interaction, multimedia communication and medical imaging. Dr. Chan was an active contributor to the Advanced Displays and Interactive Displays FedLab funded by ARL, and was a program co-chair of the DARPA-funded Multimodal Speech Recognition Workshop in June 2002.
Out of the CAVE: The Ascent of Visualization
Robin Bargar, Dean, School of Technology and Design
New York City College of Technology, CUNY

Abstract:
The 1990's ushered in a decade of "upscale visualization;" both in terms of computational prowess, advancing upon previously intractable problems, and in terms of large scale dedicated systems. Both endeavors bore fruit; some we are still trying to harvest as it ripens on the vine. With the end of the 90's the Supercomputing Meets Hollywood mentality has passed into a more diversified field of devices and users. How shall we assess our institutional position in terms of formal vs. informal training for students, as well as directions of research and investments in display systems?
Title:  Real-time Vision & Modeling
Dr. Anurag Mittal, Dr. Visvanathan Ramesh,   Siemens Corporate Research,Princeton, NJ

Abstract:

Rapid and Real Time Mapping Technologies
Mr. Jeffrey M. Young, Regional Director,  Leica Geosystems - Americas

Abstract:
Technologies available to first responders for rapid map creation are undergoing significant advances. Multi-sensor, high resolution airborne pixel-based and lidar systems are becoming more widely used as the long-standing constraints of cost, downlink delivery and spectral and spatial resolution limitations are mitigated. This discussion will include a description and system architecture for multiple use cases and a prototype of real time mapping capability. This capability is intended to support the first responder community in the event of a variety of human-induced incidents and natural disasters as well as in response planning and post incident analysis activities..

Biographical Sketch
Mr. Young has over 27 years of sales, program and project experience, including more than 15 years in senior management of GIS corporations. Mr. Young is employeed as the Regional Director Leica Geosystems - Americas located Denver, Colorado managing the activities of  regional sales persons, and overseeing  the development and operation of a network of image processing and photogrammetry software distributors. Previously, Mr. Young was Vice President, Global Solutions Sales & Marketing at Space Imaging, LLC. He also served as Executive Director, Global Solution  Sales, and as Director, North American Sales.  Prior to working at Space Imaging LLC, previous positions and assignments held were Sales Director, Sales Account Manager, GIS Sales Business Development Director, Business Unit President-Criminal Justice and Public Safety, GIS-T Program Manager, Program Manager, GIS System Consultant, Supervisor, and Office Manager. As a  Staff Scientist Mr Young has responsible for Geographic Information System (GIS) solution design and applications development, sales, business development, infrastructure management applications, facility management applications, information system design, siting studies, environmental monitoring,environmental constraints analysis, land use analysis, terrain analysis, remote sensing, computer cartography, geographic field techniques and computer training. Mr. Young resides in Centennial, CO with his wife and two children and enjoys traveling throughout the western United States.

Equinox Multimodal Image Fusion Systems
Dr. Diego Socolinsky,  Director, Research & Development, Equinox Corporation

Abstract:
The wider availability of sophisticated imaging sensors in various regions of the electromagnetic spectrum creates an opportunity for exploitation never before possible. Combinations of visible, thermal and other imaging modalities allow for unprecedented capabilities in areas ranging from target detection and recognition, biometric identification and fire control systems, to name but a few.  Equinox has an active system development program in this area, supporting a variety of mission requirements for military and civilian customers. We will showcase some examples of image fusion systems currently under development.
Questions for Our Panelists
Michael Grossberg, City College of New York

 

Imaging
(1) With digital imaging there has been an explosion in the collection images and video. We also know from our own experience how important visual information is for many tasks. Yet the use of technology does not always follow our intuition. Video phones have been available for more than two decades but never caught on. Will analysis of imaging and video ever break out of niche applications and deliver on the promise of a wide range of revolutionary applications and if so why?

Vision
(2) Computer vision is often explained in the popular press in terms of  object or even face recognition. Yet some have pointed out that object  recognition remains one of the least useful algorithms. What do you see has the most promising aspects of image understanding to applications in industry?

Visualization
(3) There was a great deal of excitement and perhaps hype surrounding  VRML and the potential for interactive 3D interfaces. Yet besides limited niches  we still see the primary tools businesses use for visualizing information are 2D graphs, maps, as well as tables. Where do you see visualization research having a  meaningful impact on our ability to processes, understand and access information?

Display
(4) There are two conflicting trends we see, larger and larger displays  and smaller and smaller displays. It seems we want enormous displays on our desktops, living rooms, and control rooms with vast amounts of information such as high quality
DVD movies and the latest news. Conversely we want small unobtrusive displays that fit on our cell phones and watches. What challenges and opportunities does this create for  visualization researchers?

Industry and Universities
(5) How does your company currently interact with university partners?

(6) What are your expectations from a partnership with researchers in a university?

(7) If you have had past university interactions and partnerships can  you highlight the kind of frustrations you faced and what you think your university partners should keep in mind?

Imaging and Visualization Future
(8) Over a medium term, what research areas do you look to universities as helping you in meeting your research objectives?

(9) There is that famous line from the movie the graduate where the  family friend advises Dustin Hoffman's character to go into "plastics" for success in business. If you were advising first year CS graduate student choosing a field in imaging or visualization what field would you point to as having the greatest potential in the long term?