BRS brings machine-learning to video
HOUSTON--The latest company to enter the video analytics fray is BRS Labs, with the BRS standing for Behavioral Recognition Systems and the technology being touted as the next generation of video analytics, whereby the programmers donÃ¢â‚¬â„¢t write the rules, the software itself does the rule-making.
Company president John Frazzini explained: Ã¢â‚¬Å“We figured out a way to take what a video camera is seeing and have it be machine readable so that we can learn an environment over time and it can create rules for what to look for. ItÃ¢â‚¬â„¢s a learning technology that provides the ability to scale in a way that hasnÃ¢â‚¬â„¢t been available to date.Ã¢â‚¬Â
Essentially, BRS hopes to solve the long-standing criticism of analytics that rules-based analytics only catch the bad guy if he breaks one of the rules.
Initially, BRS will approach large and complex projects, hoping to get a return on the $20 million in start-up fund and two and a half years of research and development that has gone into founding the company.
These kinds of projects, where there may be thousands of cameras, donÃ¢â‚¬â„¢t lend themselves to traditional video analytics, Frazzini said, because Ã¢â‚¬Å“each camera has to be programmed to each scene and what needs to be looked out for.Ã¢â‚¬Â
BRSÃ¢â‚¬â„¢s system watches a scene for a period of as long as two weeks, establishes what is Ã¢â‚¬Å“normalÃ¢â‚¬Â for that scene, and then creates rules and alerts when it sees something out of the ordinary. The learning period could be shorter for a static scene, or one like a freeway where itÃ¢â‚¬â„¢s pretty clear when something is going the wrong direction, and longer for places that play host to large crowds or a variety of activities.
Some rival analytics makers might argue that their systems are self-configuring, but Frazzini said his companyÃ¢â‚¬â„¢s use of the cognitive process, how the brain actually works, takes a video surveillance system from a place where analytics give a computer the ability to see things, to a place where Ã¢â‚¬Å“we can convert what is seen by the video camera into a machine language and cause the computer to learn.Ã¢â‚¬Â