CCTV has come along way since it was first introduced to view rocket launches back in the 1940’s, with some major step changes such as: the introduction of analog multiplexers back in the 60’s - making it possible to view several cameras on one monitor; the advent of digital video recorders (DVR’s) – so it was possible to easily rewind and review remotely and more recently affordable IP CCTV cameras enabling home owners to have their very own home security surveillance systems.
The ability for technology to automatically identify the contents of images (both photographic and video) has been an ongoing development for well over 50 years. The first arrest through detection of a stolen car using ANPR was made all the way in 1981 in the UK. The Department of Motor Vehicles in New Mexico and West Virginia were the first organisations to use image facial recognition back in 1993. It would take until 2001 for real-time facial recognition from video to become a reality. Since then, real-time facial recognition driven devices have exploded, especially in Far-East and South-East Asia. With China using facial recognition systems as a means of identity by both public organisations and private companies. For example, in hospital waiting rooms, and large housing complexes, and transportation systems and urban policing.
With the emergence of Generative AI, robotics and drone technology, increased computing power and faster broadband speeds, video surveillance is now entering a new phase in it’s journey.
The power of AI enhanced CCTV security
This new phase is one in which video surveillance AI moves from basic identification, to being able to not only identify people, but also understand their behaviours, identify security and safety risks without the need for human oversight, suggest and take corrective action and even detect people’s emotions!
Big Brother is almost upon us!
To understand the power of AI enhanced CCTV security, let’s take a simple example of an office which has secure rooms which only authorised staff can enter. The guidelines state that no one is allowed to tail gate, but tail gating always happens and no one is aware of it. Confidential information can be found on tables and on screens as these unauthorized people walk in. An AI enabled CCTV system, would be able to immediately pick-up un unauthorized person tail gating into a secure zone through facial recognition, and then video and record evidence of them picking up documents or items from the secure zone – providing real time evidence and sending alerts to security staff and management there and then – with links to the specific IP theft video clips.
The above use case, would not require anyone monitoring CCTV cameras on a screen, or someone rewinding and fast forwarding hours of video to see the entry and discover if anything was taken – it would be available immediately and stored safely for action. In theory AI enhanced CCTV would remove the need for human video monitoring altogether.
Today, these activities would be taken care of by an ARC – Alarm Receiving Centre.
What is an ARC
An ARC is a is a facility that monitors and responds to alarm signals from security systems. In simple terms it is a security monitoring centre.
The main tasks of an ARC can be broken down into three main categories, namely:
Receiving alarm signals from security systems in residential and commercial buildings
Processing the signals
Taking appropriate action to protect the property
A well established ARC receives multiple alarms, and in the digital age many ARC service providers have also expanded into the realm of cyber security.
Despite this, the main task of ARC personnel is to monitor CCTV alerts and take action based on this.
Could AI replace the need for humans to monitor CCTV
So, how reliable is AI, when compared to human capability.
Arunkumar, CEO of the Institute of Robotics and Artificial Intelligence Research Foundation (IRAIRF), has designed and deployed AI enhanced CCTV surveillance and Security Robots across Asia. He believes that AI enhanced CCTV is already capable of replacing the majority of human CCTV monitoring tasks – but not all. “AI can be trained to pick-up specific behaviours, such as a person looking at their mobile phone. But what about anomalies – there is no such thing.
“Just like with monitoring internet activity or financial transactions, AI can be setup to identify anomalies. When something is not expected. This can raise an alarm. In these cases, you would need people to look at the anomaly.
“But this is when AI can be very clever. If it something unknown for the person viewing the anomaly, AI can in fact help with steps in analysing the anomaly depending on the specific situation. Providing guidance on problem definition, and then helping to guide the person in finding the correct solution – especially when armed with the knowledge of compliance standards.”
In Arun’s opinion, we are several years away from 100% AI run CCTV solutions, especially for complex sensitive physical sites.
Another point that Arun makes, is that of video surveillance being used as legal evidence. In this case, subject matter experts would have to be involved in reviewing the video evidence, and ensuring they are used in a correct manner in court.
“By combining AI video surveillance with robotics, we took a step change in security surveillance. From stationary cameras we were able to have moving cameras, working at all times of the day, identifying and reporting incidents immediately. It also meant, that if a static CCTV camera identified a problem, then a robot could be deployed to the site of the incident immediately.
When dealing with large sites, the robots would also be able to identify specific building faults or areas that required maintenance. So, the combination of AI enhanced CCTV and robotics is very likely to be a game changer.”
More democratization of complex CCTV security operations, rather than replacement, of the ARC
There is a case that AI enhanced security can reduce a large part of the human effort of ARC personnel in monitoring and processing CCTV real-time video and raising alarms and incidents more effectively than ARC personnel can.
However there are a number of reasons why ARC’s are still required. These are as follows:
· For companies and organisations where safety and security is key, an ARC must meet compliance (BS EN 50518: Monitoring and Alarm Receiving Centre). An AI enhanced CCTV solution cannot meet the compliance alone, it has to be part of a larger operation;
· ARC’s use more than just CCTV to monitor for security;
· AI enhanced CCTV needs to be trained to identify situations – not all situations can be covered. However, the majority for a given organization can be managed.
· Taking appropriate action, especially in the case of ensuring the video evidence is permissible in court and handled accordingly is something that must be looked into.
The reality is, that AI enhanced CCTV will reduce the barriers of entry for many security companies to offer ARC services to larger clients who require more sophisticated solutions with tighter Service level Agreements.
It will also increase the number of video surveillance incidents that are raised and require action – meaning that the ARC provides a more robust security and safety service.
So, in summary, AI is likely to increase the number of security companies able to provide high end CCTV ARC services and also increase the demand for these services.
Comments