Seeing Eye Cameras

• By Michael Fickes
• 03/01/09

One night not long ago, officers in the security center of the Miami-Dade County Public School system were going about their business, which no longer included watching video surveillance monitors.

In the background, one of the surveillance monitors showed a man in a corridor tearing open plastic trash bags that had not yet been carried to the dumpster.

Suddenly, the alarm monitor flashed on. It showed the man going through the trash and even highlighted him in a box painted on the screen. The system also sounded an audible signal. A security officer looked at the monitor, checked the location of the camera, and dispatched patrolling security officers.

Within minutes, officers arrived at the scene and escorted the man, who turned out to be homeless, from the building.

The incident illustrates the capabilities of a new video surveillance tool being used in a handful of schools within the Miami-Dade County system. Called video analytics, the new tool gives video surveillance systems the ability to recognize certain shapes.

In lay terms, the system searches every frame of video from every camera connected to the system and sends alerts upon spotting people or vehicles where they should not be or engaged in activities that should be looked into.

That capability transforms video surveillance into something much more than an investigative tool used after a theft, break-in, slip-and-fall, or other breach of security or lapse in safety. Video analytics systems make security officers proactive. The technology sees potential safety or security problems as they take shape and alerts a security officer who evaluates the situation and decides to intervene or not.

In the case of the homeless intruder, the Miami-Dade County Schools Police Department programmed the system to send an alert if a person showed up in that corridor during the night. Security officers arrived before the intruder caused any damage. With a conventional system, someone would have found trash strewn around the corridor the next day and reported it. Only then, would security officers have checked the video and discovered the intruder — too late to prevent the man from making a mess.

In addition, video analytics systems that have been properly configured will detect more or less all of the events that need looking into. Human monitors cannot do that. Studies show that when even the most experienced security officers study monitors, they typically lose their focus within 20 minutes or so. Analytics never lose focus.

It’s important to note that analytics do not detect bad behavior. Instead, the technology identifies events worth checking. Replace the homeless man in the story above with a janitor setting about carrying the trash bags outside to the dumpster, and the system would still have notified the security center about an event that might bear investigating. In that case, of course, the security officer monitoring the system would have seen the janitor and canceled the alert.

“It isn’t an automated threat detection system; human beings still make the judgments about what is and isn’t a threat,” says Edward Troha, director of Global Marketing for ObjectVideo, Inc., a Reston, VA-based company that provides intelligent video analytics software to original equipment manufacturers, who embed the software into cameras, digital video recorders, servers, and a variety of devices used in the design of video surveillance systems.

How the Technology Works
Many security professionals who haven’t run into video analytics often confuse the technology with older motion-detection technology. “Motion-detection is not an intelligent technology,” Troha says. “It detects the motion of pixels in the video frame.”

If a bird flies into the frame, a motion-detection system sends an alert. If it rains, the system alerts. If an intruder climbs over a fence, alerts go out. It is a technology that sends out false alarm after false alarm and creates the risk that when a security problem does arise, officers may at first assume that it is a false alarm and won’t respond quickly enough.

By contrast, says Troha, video analytics technology uses complex mathematical formulae called algorithms to analyze the movement of pixels in a video frame. How many pixels are moving indicates size. What shape the moving pixels outline indicates what it is. The kind of motion is also evaluated.

The algorithms chiefly look for objects with the size, shape, and movement characteristics of people and vehicles. ObjectVideo analytics can recognize a person loitering, walking, running, moving through a door, and other kinds of movements involving one person. The technology can also identify people converging or the formation of a crowd.

It can identify a vehicle being parked or driving past. It will notice if the vehicle is moving at a high rate of speed. It will notice if one vehicle seems to be chasing the other.

An analytics system can be programmed to look for some or all of the available scenarios all of the time or at specific times.

How Miami-Dade County Schools Use Video Analytics
Generally, a Miami-Dade school uses video analytics technology to monitor the entrances, exits, parking areas, hallways, gym, cafeteria, and auditorium. When the system detects a person in one of those areas after hours, for instance, it sends an alert. Usually, it is a student, teacher, or administrator working late. But if it is someone the officer doesn’t know, a security officer is dispatched to check.

Two sites that receive special attention in the Miami-Dade schools are snack areas with vending machines and spaces behind the staircases.

In a couple of schools, vandals have made a sport out of knocking over the vending machines. Cameras monitoring those areas send alerts when the system sees the vending machines move. Once students came to understand that security officers would show up almost instantly when a vending machine moved, the vandalism stopped. The savings helped pay for the system.

Second, the design of a couple of schools provides hiding spaces in the stairwells. Students interested in cutting class used these hiding places between classes, waiting until the hallway cleared to go on their way.

The cameras watching the stairwells have been programmed to count the people going into the hiding place and coming out of the hiding place. When fewer people come out than go in, the system sends an alert. A security officer shows up and shoos the kids off to their classes.

“Administrators have seen a decrease in significant types of behaviors, and they have rolled out the system to a half dozen schools,” says Loye Matika, sales director with Miami-based Phoenix IVS, the systems integrator that designed and installed the Miami-Dade system.

Costs and Benefits
Miami-Dade has thousands of cameras across its 300-school system. The cameras have been installed school-by-school over a number of years. They are analog cameras connected to security stations by way of coaxial cabling. Intelligent video systems require digital signals.

To eliminate the need to buy new digital cameras, Phoenix IVS recommend installing devices called encoders on the existing cameras, continues Matika. The encoders can receive analog signals from the cameras, convert the signals into digital format, send the digital signals through a chip containing the video analytics software, and then connect to any nearby Internet Protocol (IP) network router or switcher, that then carries the signal to the security center. The ability of digital signals to travel over the existing data network offers significant cost savings — extensive coaxial cabling systems rank as one of the costliest components of an analog video system.

Miami-Dade took the recommendation and commissioned Phoenix IVS to install encoders with analytics near each of the cameras in one of the district’s largest high schools: Miami Northwestern Senior High School, with 1,700 students and 150 faculty and staff. The staff includes 10 security officers. In addition, the Miami-Dade Schools Police Department has assigned a couple of officers to the school. One hundred twenty video cameras equipped with video analytics monitor activities.

The video analytics system makes the school security staff more productive. Since no one has to monitor video anymore, officers in the security center can carry out other tasks, confident that the analytics will pick up events and activities that need to be investigated.

Outside of the security center in the corridors, classrooms, gym, cafeteria, parking lots, and grounds, the cameras add new pairs of eyes to patrolling officers. The system also deters bad behavior, especially now that students have learned that the cameras are effective.

Without doubt, proactive security stands as the most significant benefit provided by video analytics. The camera will alert, for example, when it sees a crowd forming, which could indicate a fight. An alert will flow from the system when it detects two people running. They might be running away from the scene of an incident, or one person could be chasing the other. It sends an alert when a locked door opens and a person enters. In each case, the alert draws the attention of an officer in the security center, who takes a look, evaluates the activity, and depending on what he or she finds, dispatches officers or cancels the alert and goes back to work.