The global video surveillance market is booming, up 1.9 % in 2015, 3.9 % in 2016 and up 9.3 % in 2017.
For a human, making sense of real time action reflected in an array of dozens of CCTV monitors is headache producing. It’s logical that computers can churn through data faster than we can and it makes sense to relegate mundane tasks to a system that will not get a headache, or bored and can out work humans on their best days.
Using security technology can potentially free security personnel to interact and engage, get out of the control room and into the field or, have an extra set of eyes and ears in the field..
Although CCTV has been around a long time, with every passing decade there are substantial technological improvements and enhancements. Video analytics is one such means to improving security. Currently, most video analytic systems are rules based, they answer yes or no – is there movement in Sector B? But with the addition of analytics based on artificial intelligence (AI), systems are now able to learn. Indeed, the next generation of video analytics leans on deep learning enabled recorders and servers.
Deep learning is a form of machine learning that uses a computer model inspired by human brain structure. That’s why it’s also referred to as neural networking. Neural networks and deep learning have allowed computers to come a long way. Their application include quaint examples like automatic colorizing black and white photos, pixel restoration, the ability to win computer games against humans and driving self-cars. But video analytics can offer real time insights into the behavior of vehicles, people and objects. Leverage facial and general recognition software, counting software, and the ability of a system to measure speed and you’ve got a potent tool. In this way, analytics makes video quantifiable and most importantly, actionable.
BriefCam has been offering this technology for about 10 years and according to the website, their system is able to detect, track, extract and identify people and objects from video to include differentiation of men, women, children, clothing, bags, vehicles, animals, size, color, speed, path, direction, dwell time and more.
For CCTV to work in the first place, images have to be clear. Clarity is a function of among other things, light. Shadows, glare and low light result in poor image quality which in turn results in higher false alarm rates. This is where high resolution cameras and thermal imaging can help. An optimal system deploys a broad array of approaches including thermal with VCA, visible light and infrared light to support the best possible image capture environment. But teaching systems to become familiar with data patterns can also help alleviate the false rates. This is where artificial intelligence can also step in and read between the lines, using algorithms to augment images where needed.
At this level, video analytics especially when coupled with AI can point out irregularities in human behavior and situations and provide a security officer with actionable data.
Video analytics is used in a wide range of environments – transit hubs, schools, manufacturing and government facilities; the potential application of the technology is endless. Law enforcement is using it not only to investigate incidents after the fact but also to monitor real time trends in traffic, crowd and pedestrian behaviors. In some prisons, video analytics is being used to aid in muster/roll call and for alerting prison personnel to the outbreak of fights and disturbances.
Obviously, for a CCTV system to be optimized it has to be combined with appropriate sensors and its effectiveness relies on the proper integration of all components and data inputs. For it to be efficient, security personnel have to be well training on the system’s use.
Technology is a great tool but over reliance on it guarantees failure. At the end of the day (at least for now) we humans are able to engage and assess human activity in ways that are critical to effective security. CCTV and video analytics is no substitute but rather a welcome complement to a robust security operation.