Can Video Analytics Be Justified?

Oct. 27, 2008
Intelligent video might make more financial sense than you think.

So many projects, so little money. That is the often-heard refrain of CFOs. It’s not hard to see why they feel that way, since audits have proven that the majority of corporate investments in system projects do not produce the return their sponsors promised. Still, there are some technologies that are so revolutionary, they beg to be tried. Video analytics is one of them.

The concept of having a computer “watch” your video for you is particularly appealing. We don’t need a bunch of PhDs in behavioral science to tell us that after 15 minutes of staring at the monitor, most operators would not notice Osama in their front lobby. Analytics technology is not cheap, however. How do you explain to your CFO that adding analytics to every camera in your facility could more than double your investment in CCTV? Increasingly, security directors are finding ways.

What Do We Mean by Video Analytics?
So many manufacturers are using the terms “video analytics” and “intelligent video” to cover a variety of concepts, that it is creating some confusion in the marketplace today. For our purposes, video analytics implies more than simple motion sensing. It implies analyzing and recognizing objects (people, cars, trucks, luggage) and applying predefined rules to their behavior. Rule violations such as people approaching each other on a dark street, baggage being left behind, or a truck parked too long in the same place, generate an alert that brings the situation to the operator’s attention. The power of such a system goes well beyond simple motion detection, and it is the first technology to actually deliver event-driven video surveillance, reducing the need for humans staring at screens.

Justifying Security Systems
Justifying any security system to management can be difficult. Most manufacturing or administrative IT systems get purchased by promising either cost savings or cost avoidance (productivity improvements). Security systems often add the third and more difficult factor of risk avoidance. Let’s dig deeper into all three and look at the specifics of video analytics.

The Case for Cost Savings
“You would assume many of these companies would use technology to redeploy more people from monitoring into the field. In reality, many security departments are so skinny anyway that there is nothing left to cut or redeploy. They are down to one guy, and they can’t really cut him,” commented John Szczygiel, president of Mate Intelligent Video, a producer of video analytics. While not everyone is in that situation, Szczygiel’s comments nicely sum up the problem with justifying a system through cost savings; “hard dollars” are not easy to come by. But there are still ways to make your case.

• Manpower reduction. Almost any application that reduces manpower has hard cost savings, of course. If you do have a significant number of operators staring at screens, analytics can pay for itself quickly.

• Overtime. “We are a category three airport, but our traffic peaks at category two levels. We chose to install video analytics because we are trying to automate as many security functions as possible,” said Mason Short, executive director of the Rapid City Regional Airport in South Dakota.

• Replacing guards with cameras. The classic example is the monitoring of airport concourse exit lanes. It is hard to imagine a less rewarding job than sitting on that stool, and yet the TSA has only recently begun to allow its guards to be replaced by analytics. “TSA is starting to shift the cost of exit lane monitoring to the airports. We were able to purchase these systems for all three of the concourses in our new terminal. The payback will be a no-brainer,” said Robert Ball, executive director of the Southwest Florida International Airport.

• Unattended applications. Think about other applications that could go “unattended”—monitoring unoccupied buildings, after-hours construction sites, or evening campus security. This technology can help reduce your second- and third-shift staffing requirements for many sites, while actually increasing the probability of detecting a human where he is not supposed to be.

• Investigations. Think about investigators as well as operators. In an environment with a high rate of incidents, you have an incentive to put in more cameras to increase coverage, catch the events, and record them. However, a higher number of cameras produces a larger database of video, making it harder to find the clip you need. Without video analytics to help search the database, your ability to provide video clips of an incident may actually be reduced by the installation of more cameras, or translate directly into overtime.

• Reduce existing costs. Shift your thinking from lowering security costs to reducing other existing costs. For example, no one questions the probability of retail theft loss. It is not a question of “if,” but rather “how much.” In that environment, using analytics to monitor the movement of high-value assets and alerting the operator to review each transaction will dramatically lower losses and provide a clear ROI.

Retail loss is not the only example here. “Take the case of Drexel University in urban Philadelphia. They had a problem with vandals jumping the fence and tearing up the logo and surface of the football field. By adding analytics to their video system, they were able to consistently generate alarms when anyone jumps the fence,” said Szczygiel.

Cost Avoidance
While hard dollar savings carry the most weight, cost avoidance can also be a convincing argument if there is agreement that a problem exists that must be solved. Some examples might be:

• Avoiding bringing other technologies to bear. Let’s suppose we have an existing fenced perimeter, and that barrier is being breached consistently by the local vandals. What are the options? An additional buffer fence, a buried line or fence seismic sensor, or a microwave motion detection system would all involve trenching or construction costs. Adding analytics to an existing video system is likely to provide significant savings.

• Circumventing the need to add resources. The Canadian Navy has added video analytics to an existing video surveillance system to monitor unauthorized vessels through the Straight of Juan de Fuca in British Columbia. By upgrading from a dysfunctional motion detection system to video analytics, they were able to reduce false alarming by a factor of 10. “The new system not only cut the paperwork associated with those false alarms, but it improved effectiveness and allowed them to reduce their marine patrols by 50 percent,” said Bob Burkholder, security systems executive for Siemens Building Technologies.

• Making better use of the resources you already have. “Ordinary CCTV forces you to assign a dedicated (and ineffective) person to monitor video. With analytics, you have the choice of multitasking; work with paperwork, work on dispatching and know that if an event of interest occurs, it will be brought to your attention,” said Craig Chambers, president and CEO of Cernium Corporation.

Risk Reduction
Risk reduction may be the most difficult case one can make. It is problematic largely because corporate management is often not used to thinking in terms of probabilities. You can easily estimate the realistic cost of the impact of an incident. The real issue is that management often doesn’t believe the incident will ever happen to them. Let’s look at some of the key factors that would justify a video analytics system in this environment.

• What are the odds of an incident? While this has little to do with video technology selection, it has everything to do with getting a technology purchase approved. Is this a problem that the management team believes needs to be solved? The solution lies in hard numbers: your company’s prior experience, a survey of incidents of your peers, or industry statistics if you can get them.

• What are the odds of your detecting it? This, of course, is where video analytics really shines. “We have a retail shopping mall customer that found a clip of an alleged assault with less than a minute of searching. It would have taken days to search the DVR for that video, if they could have found it at all,” said Chambers.
For some situations, such as protection of artwork, statues, mobile objects such as aircraft on the tarmac, or detecting a stalled car on a bridge, there is no other effective and flexible approach. “Most often, the payback is not so much gains in efficiency as it is effectiveness of what you are already doing, as well as doing things that you simply can’t do any other way,” said Szczygiel.

Let’s face it; staring at a monitor for hours on end is not very valuable. The odds of detecting one of the rules you decided to look for goes up dramatically when a computer is doing the monitoring.

• Is there a real risk? The other piece to the justification equation is simple. If an incident does occur, what are the losses likely to be? Do they justify the investment? “We are looking for any tools we can add to the toolbox that lower our risks. We purchased a video analytics system because it will help us do things we simply could never do before,” said Tim Bohr, surveillance director of the Foxwoods Casino Resort in Mashantucket, CT.

Beyond the Finances
As the algorithms get better and hardware gets more powerful, video analytics will continue to become less expensive and easier to justify. That’s good news, because the real story here is a technology that dramatically improves our ability to provide effective security. Rather than question how to justify video analytics capability, the limited capacity of a human to watch 10 monitors at once makes me wonder how you could justify installing a system without it.

Rich Anderson is the president of Phare Consulting, a firm providing technology and growth strategies for the security industry. A 25-year veteran of high tech electronics, Mr. Anderson previously served as the VP of Marketing for GE Security and the VP of Engineering for CASI-RUSCO. He can be reached at [email protected].