IEEE 125th Anniversary Media Event: Pattern Recognition

Humans have outstanding abilities to interpret patterns in data captured by our senses. Current machines are notoriously inept in pattern recognition abilities. Researchers in pattern recognition, such as Dr. Kasturi, are developing technologies and building products that will resolve some of these issues in areas including:

- Security: Early detection and identification of suspicious activities, authentication of persons prior to permitting access to secure facilities, automated analysis of surveillance video for abnormal patterns, automated monitoring of coastal waters.

- Quality of Life: Automated interpretation of sign language to improve communication with people with disabilities, automated system to detect and prevent collision of vehicles, algorithms for automated video analysis to enable content indexing and search.

- Healthcare: Machine learning methods to spot disease outbreaks, understanding of gene expressions leading to development of early detection and treatment of diseases, analysis of medical images, tissue classification from magnetic resonance images for improved diagnostics, interpretation of brain waves to interact with computers and prosthetics.

Dr. Kasturi's team has established a framework for performance evaluation of algorithms that detect and track text, face, and vehicles in images and video.  These algorithms are the building blocks for smart video analytic software to automatically detect unattended baggage;   uncover power lines and other hard to detect obstacles in search and surveillance missions using unmanned helicopters;   identify the person you are about to meet by analyzing the images being captured by your cell phone's camera; and much more. 

Despite all the progress so far, experts have barely scratched the surface of the potential world-changing technologies and associated economic opportunities that pattern recognition is poised to deliver in the next few decades.  In the near future, researchers hope to

- automatically search and retrieve computer information based on a description of what you are looking for;

- identify the set of genes associated with specific diseases;

- predict weather and global environmental changes quickly and accurately.