"Such size installations require video analytics to be at the very core of the video surveillance network," Kattan said. "We are seeing the winning video analytics solutions are those that focus more on being applicable to large scale deployment-in terms of the solution architecture, performance and labor associated with the installation process. Agent Vi focuses on automating most of the processes during the installation phase, saving significant installation time typically associated with manual calibration. In addition, a significant and constant improvement is made to reduce false alarms and increase overall system performance, since those become critical in the user's ability to effectively make use of a video analytics system," Kattan commented.
According to Dick Salzman CPP and vice president of Marketing for Keeneo, Potomac, Md., today's analytics are especially suited to detect security scenarios in real time, allowing the user to take instant action.
"Vendors of video analytics have been gaining better understanding of what they need to do to improve their software to reduce or avoid unwanted alarms," he continued. "In addition, users have been gaining experience with the use of video analytics and now better understand how it can best be used and what its realistic limits are," Salzman said.
A good example of the metamorphosis of the use of video analytics is its use not for security applications but for business intelligence or operation logistics applications. "The biggest use is for counting of people and for calculating related metrics about their movement," he said.
Another interesting element of video analytics is the place where the processing of the images occurs, that is, the actual analytic operation. "Some companies have their video analytics solely installed and performed in centralized or distributed servers while others have it uploaded or embedded into the camera itself," Salzman said. "There are benefits to both approaches. The best use for server-based or camera-based analytics depends on the application and its installation. An example of this is with Keeneo's Easy Count model for people line counting applications where there are versions both for server based and a version using an Axis camera with uploaded Keeneo Easy Count video analytics."
People counting is certainly a growing and respected application. But there are many different variables that integrators have to get to know in order to write a specification properly for proven results.
"Although we fall under the same category as analytics (e.g. software for quickly locating specific events in recorded video), BriefCam differentiates between Video Synopsis, which presents all events and engages the human operator in making the judgment call, and those rule-based technologies that filter out or issue alerts based on those rules," said Dror Irani, chief executive officer of BriefCam, Neve Ilan, Israel. "As such, Video Synopsis is free from false alarms, meaning that the operator sees all these events and can quickly decide what is suspicious and what is not. We don't replace the human eye or brain. We just make them hundreds of times more efficient. So, in our case, greater resistance to false alarms means developing a technology whereby these events are seen but they don't matter," he said.
Irani said security cameras are becoming more ubiquitous and more accepted. "At the same time, the price of IP cameras is dropping fast, which means more small- and medium-sized enterprises will be buying cameras and software suites that include analytics serving not only their security needs but other areas as well: customer behavior tracking, monitoring personnel and the like. Our vision is to be able to enable anyone-in the home, at work or on the go-to take just a few minutes and review hours of footage remotely from the cameras protecting their property and loved ones," Irani said.