Video surveillance cameras are a blessing to security professionals. Hundreds of cameras providing live views of critical locations on large education or healthcare campuses, manufacturing facilities, maritime ports and other large sites offer an opportunity for trained professionals monitoring the data to minimize criminal activity. 

Yet, at times, these same cameras could seem more like a curse. Watching hundreds of hours of video daily can overwhelm even the best trained and most dedicated monitoring staff. Studies have shown the accuracy of people tracking video suffers over an average eight-hour shift when viewing multiple monitors. Potentially criminal or dangerous activity may be missed. A deep review of recorded video is helpful when taking remedial actions or prosecuting offenders, but often, the initial damage cannot be undone. 

Over 20 years ago, motion detectors and other sensors (eventually embedded into cameras and servers) began event-based monitoring, creating alerts when movement is detected within a camera’s field of view. While this helped focus a staff’s attention on alarms, up to 98 percent were false due to non-threatening events such as wind-blown foliage, stray animals, weather events or changes in lighting. 

In this scenario, almost all alarms are a nuisance, yet they still require an inspection not to miss an actual event. Today’s camera- and server-based analytics often lack the sophistication to distinguish between events involving a person or vehicle and other non-threatening activities. 

However, recent artificial intelligence-based tools reduce more than 90 pecent of customer’s false alarms. Powerful cloud-based software instantly alerts only on human and vehicular activity. As with other cloud-based services, there are no servers or other equipment to install at a customer’s site. And the software is easily scalable to match the number of site cameras. 

The benefits are immediate, including fewer missed true alarms, less monitoring staff burnout and a reduction in the hefty false alarm fines many local first responder organizations charge. 

Modern event-driven software is made possible by deep learning. This AI subset is exposed to millions of images of people and vehicles and then continues to learn and increase its accuracy when linked to end-user cameras. 

And the capabilities of this software continue expanding, creating more actionable intelligence for security operations. New software can detect when camera angles are moved, the device has been tampered with or has not created alerts for more than six hours. 

Protective security is enhanced with software capable of detecting the color of clothing, vehicles and other objects. Users may choose to know when people are loitering, a crowd is forming, or there is activity around specific areas of interest such as gates and doorways. Watchlists enable users to upload photos of objects to seek and observe or ignore. 

If events do occur, deep learning technology aids with forensic analysis. An investigation may include searching for alarms containing objects of a similar appearance from all cameras or selected sites. The software also tracks an object’s movements across multiple cameras.

Monitoring cameras for hours is a demanding job, made more so by dealing with false alarms.  Event-driven software has changed dramatically with advances revolutionizing video surveillance. By relying on software to handle much of the video analysis, end users benefit from more effective and efficient security operations. 

The most frequent users of event-driven analytic software are remote video monitoring stations offering remote guarding services to augment private security guards. AI-based software is vital where potentially hundreds of clients are being monitored in one central station. Enterprise and government organizations conducting their own monitoring also count on proven analytic software to better protect their employees and facilities.