As technology takes over more aspects of our personal and professional lives, companies are increasingly turning to automation to streamline simple, manual tasks and in turn, implementing advanced technology such as artificial intelligence to offload decision-making processes. The aim, of course, being to help them save the time and cost associated with having staff carry out these activities manually.
As technology takes over more aspects of our personal and professional lives, companies are increasingly turning to automation to streamline simple, manual tasks and in turn, implementing advanced technology such as artificial intelligence to offload decision-making processes. The aim, of course, being to help them save the time and cost associated with having staff carry out these activities manually.
This scenario is particularly apt in light of the current COVID-19 crisis compelling most workers to stay in their homes. With business operations hindered and productivity jeopardised, technologies like automation and artificial intelligence obviously spring to mind as potential solutions.
In the long-term, a situation where human workers aren’t needed and machines take care of everything could be appealing. However, this is simply not an option when it comes to security. While AI can provide security staff with invaluable help, eliminating human decision-making completely could cause serious incidents – and with just the technology to blame, when accidents do happen, this may put a hard stop to the exciting growth of AI in security and the business benefits it can bring. That’s why the key to an efficient, accurate and cost effective video surveillance strategy is a combination of sophisticated technology and human interpretation.
Let’s take a look at how this synergy of artificial intelligence and human input makes for a winning security approach.
Deep Learning for enhanced accuracy
Video analytics undoubtedly revolutionised video surveillance. The shift from a situation where a staff member had to constantly monitor security footage to spot potential intruders, to one where the system itself was able to alert users of suspicious behaviour was game-changing. Before, if an intrusion was missed, resulting in a break-in or damage, the footage had to be watched retrospectively to identify the offenders, with very little scope for success. Now, however, security operators are able to act in real time, while the incident is taking place, increasing their chances of preventing crimes or containing their consequences.
However, monitoring systems, just like individuals, can make mistakes. Tasking technology with flagging potential security threats meant that not all the alerts created were real alarms and they would often call security managers’ attention when there wasn’t a real need. False alarms not only waste monitoring staff’s time, hindering operational efficiency, but also impact their ability to identify anomalous events as they cause a kind of video blindness, making them almost jaded to the excessive number of alerts raised by the system.
It’s a problem of accuracy which, luckily, can be circumvented with the help of a subset of AI – Deep Learning. While video analytics powers the detection of events, Deep Learning can play a key role in enhancing the precision of such detection. A Deep Learning filter can be used to pre-train the system to only flag the presence or movement of humans and vehicles – events, in short, that can represent a security threat. This means wildlife or environmental conditions – which are the main causes of false alarms – would not trigger an alert. This is particularly useful for perimeter protection in sterile zones, where human footfall is very limited and elements like tree branches blowing in the wind or heavy rain could simulate an alarm in a conventional system and alert staff of a potential intruder – causing them to interrupt more important activities. When the alert is generated, that’s where human intervention is incredibly valuable. The AI-powered video analytics may spot an intruder, but it’s still the worker’s responsibility to review the situation brought to their attention by the system and determine how to respond to it.
As it often happens with technology, the hype around artificial intelligence is causing organisations across all fields to consider its implementation, fearing getting left behind and wanting to innovate the way they do business. It’s important, however, to evaluate how such technologies can improve their operations, before rushing to invest in them. When it comes to security and specifically, video surveillance, Deep Learning can provide unprecedented accuracy, minimising false alarms and ultimately, helping security managers and their teams protect premises like never before.
© Technews Publishing (Pty) Ltd. | All Rights Reserved.
printer friendly version