• September 1
• September 2
• September 3
• Accepted Papers

• Program Areas
• Object Detection & Recognition
• Tracking
• Activity Monitoring
• Signal Surveillance/Processing
• Applications/Systems/Tools
• Sensor Networks/Fusion/
     Infrastructure
• Performance Evaluation

Plenary Talk

Eroding or Opening?

Speaker:

Terrance E. Boult El Pomar Professor of Innovation and Security University of Colorado at Colorado Springs and CEO/CTO Securics Inc.

Automated Video Surveillance has move from academic operating on small video streams with component level algorithms into successful company products. As has happened in many of the electronics areas, leading companies in surveillance have more funding, more Ph.D.s, more data and more facilities, than even the best university research labs. As a maturing subfield, its natural to ask if the opportunities for research in surveillance is eroding there are there openings left? Well, as any vision/image process student knows, erosion is the first step of opening. This talk will argue there are many openings, and that they exists at every level of surveillance problems: from sensors to low-level processing to high-level analysis. We highlight a few very different directions where we see more open space ahead.
The first part of the talk briefly discusses the the Nobel prize winning Economic theory of asymmetric information, the "market for lemons" and Kerckhoffs' principles for security. The economic theory tell us the real answer to if the field will be eroding or opening depends on how we address these issues. The talk then discusses the role of academic, and industrial research, in light of these issue and gowning need to improve quantitative evaluation standards and call for a more formal set of benchmarks, and possibly standards for surveillance systems.
The second part of the talk addressees another view where we believe there is significant growth potential: balancing surveillance with privacy. The talk briefly points out some of the excellent work from others in this area and reviews a little of our privacy-enhancing camera work. It wil include issues in biometric surveillance.
The last part of the talk will talk about the next sensor/architecture revolution in surveillance, multi-mega pixel or gigapixel smart sensors. We exaplin why we believe these paridigm shits are inevitable. While the major persistent surveillance programs, because of the large size and potentially sensitive data, may be dominnated by large industrial contractors, we believe there are still plenty of openings for both fundamental and applied research. The talk will also discusses a bit of what you'll need to know play and maybe to help recognize the opportunities when they are staring you in the face. We give example of how the new paridigm makes some old ideas/problem, with a twist, new again.

Back to program