Today at the ASIS International show, I was struck by two things: (1) the fierce competition in video analytics and (2) the IP revolution finally emerging. Even compared to just 6 months ago at the ISC West show, the landscape really seems to have changed in these two related fields.
"Bandwidth is not a problem," said Bruce Doneff at the DVTel booth. He explained that the word (and proof) is finally getting out that not only are system designers figuring out ways to more efficiently process data so as to minimize the load on the network, the networks themselves are becoming so robust that they can handle many of the loads that were problematic a few years ago.
Doneff says that DVTel predicted a few years ago that "everything" was going onto the network and began pushing for open standards (despite resistance from others). In his words, "We were IP before IP was cool."
One specific application of IP that has improved over the past six months is video analytics. Video by itself has plenty of data to send over a network, but then adding complicated algorithms to detect "suspicious activity" even tougher. However, these past few months have made a difference. However, companies such as Agent Vi, DVTel, OnSSI, and Sony are working on ways to transmit video analytic data over a network efficiently.
And video analytics itself is improving--the algorithms are getting better. For instance, Vidient, the makers of the SmartCatch video analytic software, are demonstrating their tailgating feature live for the first time. (Tailgating is when one person enters a door closely followed by at least one other... this could be suspicious in the sense that the first person is authorized to enter but the second may not be authorized.) At ISC West, Vidient's tailgating algorithm was working consistently enough for them to advertise live demos of it. At ASIS, they're happy to do so--and claim to be the only ones.
Joe Montalbo, president and CEO of Pixim, says that video analytics industry is "state of the art," which he quips as meaning "it works half of the time." Montalbo says that often when a video analytic algorithm doesn't function properly it is simply because the video feed it got wasn't high enough quality-the natural plug for his own product, a chip/sensor that he says is one of the best for cameras that are going to be used with analytics (the company's chips can be found in a number of cameras on the show floors, including ioimage, a camera/analytics company).
So, with cameras getting higher quality images, networks becoming more efficient, and algorithms becoming smarter, video analytics is fast approaching the mainstream.
[Greg McConnell is assistant editor for Security Dealer, and regularly maintains a blog covering issues facing the dealer community.]