While BRS Labs is opposed to disclosing technical information, their patent applications do provide information on their approach. Key documents include:
While BRS Labs's marketing material often mix a behavioral approach with a rules based approach, their patents are clearer about the fundamental technology. Specifically, their patents focus on building models of historical activities in a scene and then determining the statistically similarity of new events to historical ones. This provides them flexibility to alert against 'out of the ordinary' events that rules based approaches would not trigger.
As we commented in our original report on BRS Labs
, identifying things that do not match historical conditions could provide a very useful way to spot on activities that cannot be clearly defined as rules. A shopping mall is an ideal example of this. Many incidents occur in crowded public parts of the mall where rules analytics (motion, tripwire, crowds, etc.) would be worthless as alerts would be triggered almost continuously. If a video analytic technology could alert based on historical patterns, this might make security guards more efficient in where they look. While many alerts would not be of incidents, it would likely be better than simply staring at a wall of monitors.
On the other hand, the core technical approach proposed in the patents is not revolutionary and is potentially inferior for well defined rules (like alerts on crossing a fenceline) that are the core of today's video analytic business.
Given BRS Labs significant burn rate
and the limits on applications that can benefit from their core 'behavorial approach', their challenges in winning deals and their sales push for traditional video analytic applications is not surprising.