- Security TWENTY
- Women in Security Awards
Kevin Waterhouse, Managing Director at VCA Technology, an AI-based video analytics product company, covers how ‘applied intelligence’ can drive tangible results in physical security.
With new technologies constantly emerging, it’s fascinating to watch the security space evolve and leverage them to achieve better outcomes. One of these – artificial intelligence (AI) – has taken the business world by storm; and while the hype around this technology may have surpassed its true potential for now, the possible applications in security are undoubtedly exciting.
While many automatically think of facial recognition when prompted to talk about AI in security, the truth is this tool’s real value in business protection lies elsewhere. Machine learning, a subset of AI, has the power to help surveillance technology reach unprecedented levels of accuracy – making security staff’s life easier, significantly improving protection for end-users and creating better business opportunities for the middle man.
Accuracy of detection
Historically, the main concern with surveillance applications that used video analytics to generate alerts was that they may be unable to distinguish a human from, for example, a form of wildlife – subsequently creating false alarms which waste time and resources. That’s a challenge machine learning can help to overcome, as it enables the system to be pre-calibrated to detect real threats and disregard false ones. In most security-based applications, the user merely wants to identify a person or a vehicle, both of which could represent a security threat. When it powers video analytics, the machine learning tool enables the developer to instruct the algorithm to pick up on specific characteristics and objects, similarly to how a human would visually disseminate a scene. This enhanced precision means monitoring staff’s time isn’t wasted with unnecessary alerts caused by objects or environmental fluctuations, meaning their productivity and their attention span increase – and their performance improves.
It’s also clear how the ability to trigger meaningful alarms can drive real-world benefits in terms of securing a perimeter. Machine learning-enabled analytics can detect suspicious events in real time, critically improving facility protection by empowering staff to proactively address a current incident, rather than reviewing a past one – in which case, often, very little can be done.
While technologies like AI and automation are revolutionising the way organisations function – enabling them to do more with less – business leaders are fooling themselves if they think they’ll soon be able to eliminate the need for workers entirely and cut the related costs. Of course, we are increasingly relying on machines to carry out manual tasks and even make small decisions for us – like deciding whether the shape in front of our surveillance camera is a person or a tree branch, for instance. But, in a sector like security, where a business’s livelihood or, at times, people’s lives are at stake, the value of the human input remains untouchable.
The machine learning component is no doubt of great help for monitoring teams – overworked and understaffed as they are – because it filters through the potential alerts, blocking out those that don’t meet the desired criteria (as they aren’t people or vehicles). Sure, this leaves staff members with only a handful of unusual situations to decipher. But the responsibility to evaluate the alarms, when they do come up, still lies with them. Is it a delivery worker approaching, or could it be a burglar? What is the next step? This is where human intelligence will always be more valuable than AI.
Successful businesses, both inside and outside the security sector, are the ones who manage to combine the best of technology with the best of human intuition.