Artificial intelligence in perimeter protection

How artificial intelligence is transforming the security industry

Networking and digitalisation have ushered in a new era in the security industry. Cameras and sensors are becoming ever more sensitive and are able to cover large areas. However, the increasing amounts of data that are being collected also bring new challenges: they can hardly be analysed by humans alone. Artificial intelligence (AI) is providing support here. It helps perimeter protection with its ability to analyse complex data patterns and identify potential threats in real time. This technological progress is particularly important for critical infrastructures such as airports, power plants, industrial sites and military facilities.

There, for example, AI-supported video surveillance ensures more precise and efficient hazard defence by learning from visual data and making independent decisions. „AI can also be used to enable effective real-time analysis of people and objects, informing security personnel of security-related events as and when they occur,“ explains Benjamin Körner, Architect & Engineering Manager at Axis Communications. Especially in the area of perimeter protection, AI-supported solutions can be used to recognise suspicious activities in good time to ensure timely intervention. „For example, thermal imaging cameras with integrated video analysis functions can provide reliable and space-saving protection for extensive perimeter zones. In these cases, the automatic triggering of alarms enables comprehensive protection without the need for security personnel to be permanently tied up in the security centre,“ continues Körner.

Fewer false alarms thanks to AI

Traditional monitoring systems quickly reach their limits with environmental influences such as light reflections, flying insects or wild animals and generate false alarms. „Without AI, an alarm can be triggered by shadows or animals in the image section. With the help of AI, objects can be detected and classified accordingly – for example people, vehicles or bicycles – to prevent this,“ emphasises Körner.

False alarms are a source of risk that should not be underestimated. This is because they can significantly impair the vigilance of security personnel, which could have fatal consequences in an emergency if actual events are not recognised as such. „Many unwanted alarms in a property lead to parts of the system being switched off or alarms being ignored,“ warns Wilfried Joswig, Managing Director of the German Association for Security Technology (Verband für Sicherheitstechnik e. V.).

As perimeter security takes place outdoors and the sensors are therefore exposed to the elements, the biggest challenge in perimeter security is reducing unwanted alarms. According to Wilfried Joswig, a large number of parameters must be taken into account for this – in addition to the seasons, these include the respective weather conditions, time of day, animals, people, vehicles, etc. „All of these influences must not lead to alarms being triggered and must be eliminated or suppressed as an alarm source using suitable measures. This is only possible with intelligent detectors that take these environmental influences and disturbance variables into account and adapt their response behaviour to the current weather conditions“, says Joswig. According to him, these disruptive influences are not only to be expected individually, but also in any combination and manifestation: „A storm in autumn, for example, has different characteristics than in winter.“ Compensating for these disruptive influences is therefore a very complex task that can only be realised using artificial intelligence. AI is used to evaluate the multitude of information and events and, thanks to the self-learning detectors, leads to optimised response behaviour even under adverse environmental influences and the number of unwanted alarms is significantly reduced.

AI-based algorithms score points here with their ability to precisely classify objects so that only relevant alarms are forwarded. This allows human resources to be utilised optimally and security objectives to be achieved more efficiently. „Unwanted alarms can be reduced through the use of AI. Therefore, the greatest benefit of AI in perimeter protection clearly lies in the reduction of unwanted alarms“, summarises Joswig. „This applies to all areas of perimeter protection, such as fences, video systems and drone detection.“

Unbeatable when it comes to recognising patterns 

However, AI algorithms are also particularly well suited to detecting unusual behaviour patterns or suspicious activities such as loitering and triggering an alarm if necessary. „For example, a thermal imaging camera with integrated AI can detect suspicious persons at a very early stage and classify them as people“, says Benjamin Körner. „In combination with an optical PTZ camera (pan-tilt-zoom), security personnel can then quickly and effectively check the situation in the next step and initiate countermeasures if necessary. Automatic reactions such as playing pre-recorded audio messages via horn loudspeakers to achieve a deterrent effect are also possible and provide effective support for perimeter protection.“

Technology in this area has developed rapidly in recent years. This is also due to the greater computing power available in today's perimeter protection solutions. „For example, where in the past additional computing power in the form of a PC was required to analyse image material, today this can also be done by the camera itself“, explains Körner. This means that threats can be recognised faster and more precisely and the response to security-relevant incidents takes less time.

Thanks to their ability to recognise suspicious activity and automatically trigger an alarm immediately, AI-supported systems are becoming valuable components of the security chain. „Companies achieve the best effectiveness and efficiency in protecting their property by using different camera systems in combination with AI-based perimeter protection technologies“, emphasises Körner. A current example of this is the use of security radars with integrated analysis functions. This allows even extensive areas to be captured and protected particularly cost-effectively.

However, Körner also points out that the collection of data is accompanied by a responsibility in the area of data protection. „The data collected by the camera sensor can be recorded and managed as metadata. This allows security providers to ensure that personal data is only used for authorised purposes and that the privacy of those affected remains protected“, says the expert from Axis Communications.