Artificial intelligence for surveillance: how video analysis works

Companies are increasingly investing in cybersecurity, but the cloud perimeter is not the only barrier they need to protect to keep their business safe: technology can improve the defence of physical facilities as well. Artificial intelligence, in fact, allows video surveillance to evolve into video analysis: your security cameras not only serve to ‘document’ an intrusion but can also prevent it.

Tools such as Axitea’s AI Video Solution distinguish themselves from traditional video surveillance by harnessing artificial intelligence on existing systems, providing new security solutions that would be unthinkable without AI.

Video analysis: artificial intelligence at the service of physical security

Even those who have never been inside a corporate security operations centre will have an idea of how they work from the many films in which they feature: a wall of screens to follow the feeds from the various cameras, a couple of agents with radios, ready to report anything suspicious. In films, spies and thieves usually have to circumvent the surveillance cameras, and need only play a few tricks to distract the guards in order to sneak into the bank, museum or company.

It would be near impossible to make almost any of these films if video surveillance were based on artificial intelligence. Video analysis allows images to be analysed autonomously, detecting a potential danger and alerting the operator. In this way, nothing goes unnoticed.

This type of solution also reduces false positives: surveillance becomes not only safer but also more efficient, not raising alarms when they are not needed. By combining the algorithms of the machines with the experience of the operators, the entire defence system becomes more efficient.

No new hardware is needed

Another classic film cliché is to see futuristic technologies designed to protect access to corporate sites or bank vaults. Pressure-sensitive floors, laser beam grating, and who knows what else. The kind of equipment that only Mission Impossible or James Bond films can afford.

Video analysis, however, does not require this much outlay for sophisticated hardware. For example, you can apply Axitea’s AI Video Solution to any video surveillance system you already have in place, as it can analyse video streams from any camera or device, new or existing.

It is not the international spy hardware that makes the difference, but the intelligent algorithms that Axitea has developed that increase the security of SMEs as much as large enterprises.

Video analysis for security: the added value of artificial intelligence

Implementing AI in a video surveillance system is therefore simple and does not require the system to be revolutionised, but once it is installed, it can recognise events automatically. Not only does it detect with a level of attention that no human could guarantee, but it optimises the video images shot to reduce false alarms. Solutions such as Axitea’s allow the user to customise the analysis, meeting specific requirements based on the type of company, the type of environment, and the type of risk.

The artificial intelligence has received training that no operator can ever receive: the amount of data that has ‘trained’ it is enormous, but unlike a simple motion sensor that would create a false alarm for every cat crossing the yard in front of the company, the programmers have taught it to look at events in a similar way to how an experienced operator would. It can detect objects and people in complex scenarios, for example it can recognise a vehicle and classify it by type: car, motorbike, bus, van and also in complex environmental conditions: when it is raining heavily or if the wind moves a branch into the middle of the view. Alerting the SOCs (Security Operations Centres) when needed, without creating false alarms and without letting intruders escape.

Not only anti-intrusion

The ability to assess any type of intrusion into the company perimeter, classifying it precisely and alerting security only when needed, is a huge strength of video analytics, but it is not the only way artificial intelligence can be applied to the use of surveillance cameras.

In addition to detecting intruders, with solutions such as Axitea’s, objects, people and vehicles can be counted and useful security information can then be extrapolated. The same goes for the recognition of ‘loitering’ near the area that the system is to protect: AI recognises this kind of attitude and helps operators to understand whether it could be a sign of impending intrusion. It also allows the user to set ‘Privacy Masks’, so that their surveillance videos respect all the legal parameters and do not cause problems.

They can also be used to assess road safety problems such as prohibited parking, traffic congestion or the people transiting through sections in which the vehicles run. And again, they can detect smoke and flames, becoming an integral part of the fire-fighting system.

With a tool like Axitea’s AI Video Solution, your old cameras not only become intelligent, but multiply their functionality.

Video analysis and security: for best results, combine artificial and human intelligence

The power of artificial intelligence for video analytics is obvious: they make surveillance more effective, more efficient, better. Detecting an intruder becomes much quicker, with the possibility of generating pre-alarms before the suspicious individual even enters the building. They drastically reduce the number of false alarms and offer extra services that go beyond simple intruder detection.

But as Axitea clearly explains, even the best technologies need the experience of those who have always worked in this field. In fact, the company offers a managed service: when the AI detects a potentially dangerous event, it alerts the SOC (Security Operation Centre) where qualified people analyse the images. In this way, they can intervene immediately by alerting field workers or emergency services, depending on the situation. Technology needs people to make a difference.

Source: Article by Tech Business