The Realities of Video Analytics: An ISC West 2007 Report

March 29, 2007
Vendors and integrators at ISC West 2007 speak candidly about the challenges of this emerging technology

Can you tell a customer "No" the right way? If so, you may not want to get involved in video analytics yet. That was one of many points that a number of vendors touting video analytics solutions made today as I bounced between product demonstrations, a Verint panel on analytics and press conferences.

It's tough to pin down video analytics -- that at least is a fact of life. It's a technology that really was "future talk" three years ago as far as most of us were concerned. Now, in March 2007, in the middle of the ISC West security tradeshow, it's clear the technology has come quite a ways. You can mark the recent Honeywell acquisition of ActivEye as a key indicator, since it's been a truism in the security industry that consolidation doesn't usually begin to occur until technologies are starting to mature and leave the "bleeding edge" environment.

I spent a good bit of my day chasing down this video analytics trend, speaking with those on the front edge of the deployment and integration of this technology. That led me to the likes of Carolyn Ramsey, director of program management with Honeywell's Automation and Control Solutions group; Roni Kass, CEO, ioimage; Chris Milite, manager of security and privacy technologies for IBM; Steven Russo, director of security and privacy technologies for IBM; Chris Taylor of Verint; and Melkon Babigian, CEO of IK Systems.

By and far, this group of individuals, which mixes integrators, vendors, and engineers, is seriously excited about what analytics means for the security industry. There were, I believe, some universal points that all of these thought leaders shared as they demoed their video analytics systems for me on day one of ISC West 2007, and I have tried to distill some of their tips, tactics and trends as we all try to get our hands around this paradigm shift in how to use video.

Tips, Tactics, Truisms and Trends of Video Analytics:

1. It's not just for security. This was THE universal point that all agreed upon. Analytics may be wonderful for setting up perimeter security, but don't neglect the applications that are core to business operations like tracking how customers move through a store or counting customers.

2. You can't set up cameras the same way you always did. You're probably not going to have much luck trying to count customers unless your cameras are directly overhead. Similarly, in applications like perimeter security, you may need to rethink camera placement as well.

3. Be ready to tell customers "No, it can't happen." There is a lot of hype about analytics and a lot of over promising. End-users may be coming into this area thinking some analytics are possible and ready today. Most who I spoke with agreed that they could do almost anything, but that end-users need to be educated on the difference between what can be offered off-the-shelf, and what would require a team of algorithm engineers to spend a month developing.

4. Analytics could use a bit of benchmarking. This was a point I actually raised to industry analyst Steve Hunt (formerly of 4A). Steve is thinking along the same lines, and I'm also hearing from many vendors that they're ready for it, and that it would assist both themselves and their customers separate the pros from the "all talk" vendors.

5. You can sell the efficiency, not just the security. ROI has been a big part of the analytics sales drive, whether that's creating efficiencies by moving guard staffs from staring at monitors to being able to respond to events, or simply being able to use event triggers to manage bandwidth needs.

6. There's a place for both analytics-at-the-edge and analytics-at-the server. What do I mean? If you want to lighten the hardware load and the centralized processing required, analytics embedded in cameras and encoders is a great solution. But don't forget that analytics can also be applied to existing forensic video during investigations.

7. Analytics at the edge is useful, but be wary of deleting all that video except for the clips associated with event triggers. If you think you can deploy an intelligence-embedded encoder to decide what video to store, remember that sometimes you may want complete archived video to look for a known previous situation that wouldn't have set off any existing triggers. You can't re-analyze the video you captured if you didn't save it, right?

8. False alarms aren't just the bane of guys who install burglar and fire alarm panels. Video analytics can generate its fair share as well, which is why integrators need to pay close attention to whether a solution is right for a customer (see my earlier point about being ready to say "No").

9. Processing power is the current challenge to doing more with analytics. These are complex algorithms that aren't going to necessarily run on any chip.

10. Analytics is more and more ready for off-the-shelf deployment. Gone are the days when if you wanted analytics set up on site, you'd need to hire 3,000 hours of a vendor's engineering crew to get it done. From server-based platforms that can be set up in 10 minutes to cameras that are practically plug-and-play analytics device, to software that ready to go and in the can, you don't have to develop everything individually anymore.

11. Camera tampering is the perfect entryway to video analytics. Put together analytics for a common occurrence, and help identify whether it's your cleaning crew who is bumping your cameras out of place, or if it's a malicious act.

12. It's taken until 2007 before the hardware was ready for going on the IT network. With much of the hardware and software for video analytics being designed to operate on the business network, integrators and corporate IT security staff were very mindful of whether the equipment met standard IT network protocols (for traffic bandwidth, virus protection, etc.). Unfortunately, until very recently, much of the equipment wouldn't have passed a standard IT security check.

If anything, the spring 2007 market for video analytics is shaping up to dispel some of the myths and misinformation that have been spread. On top of that, I sense that there seems to be a bit of refreshing honesty occurring in the video analytics market. And seeing that honesty just seems to make video analytics much more realistic, much more human, and much more ready to deploy in a sometimes dishonest, crime-filled world where tired eyes can't watch all of the cameras.