In the 2002 movie “Minority Report,” the government's Precrime Unit identifies crimes before they are committed and jails the would-be offenders. That's far-fetched, but an emerging field of surveillance techniques called video content analytics (VCA) comes close to it by predicting crimes based on a person's movements — all without human intervention.
Viseum’s “Bug” features nine cameras: Eight peer in every direction and, when the system identifies a predefined suspicious activity such as loitering, the ninth pans in that direction and zooms in. Viseum is one example of video content analytics, a next-generation surveillance technique that identifies crimes based on a person’s movements and without human intervention.
Credit: Courtesy of Viseum
For example, suppose a surveillance camera that monitors a city street spots someone standing near the curb. Cars frequently pull up to the person, stay a few moments, and then pull away. Is that person a drug dealer, or a helpful bystander giving directions? A human monitoring the camera could distinguish between the two, but a human also requires a salary and health insurance, driving up the cost of protecting an area.
VCA aims to eliminate — or at least reduce —the need for a human monitor by using information about an area and sophisticated computer algorithms to make those distinctions. AV pros who target the security market should watch this technology; it reduces a client's personnel costs, and should make video surveillance more affordable for companies and municipalities.
VCA systems also can improve security by watching a larger area at once without a comparable increase in personnel, and instead of constant camera panning in the hope of catching a crime the moment it happens in that particular direction. The endless panning, often referred to as “preset tours,” essentially is a crapshoot from a security perspective. The longer each scan takes and the wider an area it encompasses, the less likely the system is to record useful information or catch a crime in real time. VCA vendors say the technology eliminates the need for preset tours.
“Everyone in the United Kingdom uses and makes do with preset tours,” says Stuart Thompson, managing director of Viseum Limited, a U.K. company that makes VCA surveillance systems. “The police who have used and benefited from evidence [captured by the] Viseum-driven systems have remarked very encouragingly because of the extended evidence from the recordings.”
Stand Out from the Crowd
Video surveillance is a growth market. For example, the hardware portion — including cameras, digital recorders, and video servers — will grow from $6.6 billion in 2006 to $11.9 billion in 2011, according to iSuppli, an El Segundo, Calif., analyst firm.
VCA is a nascent part of the video surveillance market, but it's projected to grow more than twice as fast. In 2006, worldwide VCA systems sales totaled less than $100 million, according to iSuppli. By 2011, the firm forecasts sales of about $450 million. That outlook helps explain why the VCA market is already getting crowded.
Among the other VCA competitors are Object Video, 3VR, and Vidient, says Mark Kirstein, vice president of iSuppli's multimedia content and services practice. “A lot of the systems vendors also have VCA,” he says. “For example, GE Security recently spun off its Interlogics group into a private venture.” As a result, each vendor is looking for ways to stand out from the pack and compete on something other than price.
“Viseum appears to have the whole solution put together, with recognition, tracking, and response all integrated,” says Kirstein. It has a “very compelling value proposition, especially for advanced installations.”
Who's the Bad Guy?
A VCA system's value depends largely on its builtin intelligence and on the information it is fed. For example, to determine whether a person on the street is dealing drugs or giving directions, the system would have to know what types of activities go on in that area. It could be near an arena that hosts events that draw suburbanites who are unfamiliar with downtown. It could have a reputation as a good place to score drugs. It is vital to have all this contextual information about a particular area under surveillance.
But there's a catch: Most environments aren't that clear cut. For example, suppose the surveillance system detects people dawdling outside of a row of storefronts. Are they loitering or window-shopping? If most of them are window-shopping, can the system pick out those who aren't, who are potential pickpockets or muggers? These types of questions are important, because they affect whether they system triggers alerts that dispatch police to a crime in progress, or just waste resources by sending them on a wild-goose chase.
Eventually, VCA systems could lower insurance costs.
“The concept of Viseum will obviously reduce insurance premiums,” says Thompson. “However, the technology is too new to be recognized by insurance companies at present.”
Even when the system has ample contextual information about an area, it also needs algorithms and other intelligence to separate the good guys from the bad guys. The level of sophistication is one way VCA vendors can differentiate their products from each other's offerings. Viseum's system, for instance, uses a set of profiles based on size, shape, time, speed, and location. If a person moves in a way that matches one of those profiles, the camera zooms in and can be configured to issue an alert.
A hypothetical example helps explain how those patterns work. Suppose the Viseum system detects a person walking down a sidewalk, and that person suddenly stops, looks around, and then resumes walking. “If this stopping and starting occurs three times or more, this will be treated as a potential problem, because studies have shown that this is how somebody behaves before committing a crime,” says Thompson.
Other unusual behavior that a VCA system can flag as suspicious includes:
- Staggering, which could indicate drunkenness or severe injury, such as a person who was just stabbed.
- Someone running in an area where everyone else is walking, or going up the down escalator. In both cases, that behavior could indicate the person is fleeing a crime scene.
- Someone walking into a store with nothing and running out with multiple items. That behavior indicates possible shoplifting.
One challenge for Viseum as well as its rivals is how environments can change during the course of a day. “There were many technical obstacles around the mathematics necessary to guide the moving camera accurately, as well as many obstacles around detecting objects at distance, within line of sight, night and day, in all weathers,” says Thompson. “We are aware of many businesses that have taken on development of this technology, but they have given up after several years of trying because of the obstacles that we have overcome.”
One advantage to Viseum was the availability of faster computer processors that could run the sophisticated processes necessary to weed out criminal activity from everyday events. Another is that Thompson and his colleagues were willing to plug away at the problems. “The main reason that we had overcome all the technical barriers was the personal brute determination of the founder engineers,” says Thompson.
Making the Sale
Viseum sells its products through a variety of channels, including security integrators that resell it and by licensing it to vendors that build the technology into their own video surveillance systems. The business case for VCA systems such as Viseum's depends partly on their ability to reduce the number of people needed, such as those monitoring video feeds from cameras. There are also potential intangible benefits. For example, a shopping center could publicize the system in order to attract consumers who might associate the area with crime. More shoppers means more sales.
Tim Kridel is a Columbia, Mo., freelance writer and analyst who covers telecom and technology. He can be reached attimkridel@mchsi.com.