Jon Squires, Head of Technology Partnerships and Enterprise Solutions at Synectics, examines the role Artificial Intelligence (AI) plays in public space surveillance.

Public space surveillance requires significant resources.

Operators continually monitoring feeds from hundreds of cameras spread across our towns and cities can easily suffer from observational fatigue, leading to missed incidents or slow response times and seriously impacting people’s safety.   

AI-driven video analytics offers a straightforward solution by automating risk detection and dramatically speeding up surveillance footage interrogation.

It is already being leveraged in many public space settings for threat detection on the London Underground. 

The benefits of AI are too great to ignore and with careful management, security teams can maximise the help AI can offer in surveillance.   

As with any new technology, there can be a knowledge gap and uncertainty.

Quite rightly, authorities have concerns about specific applications, how best to implement AI and the legal implications – particularly in relation to public privacy.   

Protecting the public and their privacy  

Safeguarding privacy is always a priority for public space surveillance teams, especially when AI is concerned.

Authorities want to protect citizens in all aspects, including their right to privacy.  

Therefore, using AI-based analytics, such as facial recognition, requires scrutiny.

Doing so highlights some interesting points.   

Firstly, it’s wrong to assume that the public thinks facial recognition is fundamentally a bad idea.

Research by the Ada Lovelace Institute has found that public acceptance of facial recognition varies depending on the purpose and organisation involved.

For example, while 71% of people believe the police should be able to use facial recognition in public spaces if it helps to reduce crime, only 23% are comfortable with shops using it to track customers.  

This shows that application matters. If AI is used appropriately and ethically, it can and does have support, which is interesting because AI can actually help here, too.   

For example, facial recognition analytics that match individuals present in footage with watch list data can also automate the redaction of all other faces in the shot.   

The principle doesn’t just apply to faces.

Depending on the requirement, AI can also be trained to spot and redact (blurring or completely covering) a whole host of details from video footage, such as vehicles, building numbers and license plates.   

What this also means is that video required as part of police or insurance investigations, or for Data Subject Access Requests (DSARs) and Freedom of Information Act enquiries (FOIAs), can be securely shared with third parties, while still protecting the privacy of individuals features in the footage.   

The legal implications of AI in public spaces  

Another concern organisations have regarding AI is the lack of legislation in the UK.   

Traditionally, the Government has taken a light-touch approach to encouraging innovation.

But given the likelihood of more regulations being implemented, it’s understandable that operators are hesitant about investing in AI technology when they don’t know exactly what’s around the corner.  

Again, the key here is not to be hindered by ‘what ifs’ but to look at what’s real.

Legislation will come, so looking at surveillance solutions that allow AI to be adopted flexibly is a practical solution.

Cameras and control software that enables AI-based functionality to be easily turned off/on or enhanced through additional integrations means it isn’t necessary to wait and see.

Instead, it’s possible to implement and adapt.    

The very real benefits of artificial intelligence   

It’s much easier to appreciate the potential benefits with fresh enthusiasm by looking at how concerns around AI can be addressed.

The primary reason for public space surveillance is citizen safety and AI-driven video surveillance has many applications which support this.   

AI can be trained to spot scenarios indicative of risk proactively.

For instance, it can easily recognise and categorise different combinations of shapes, such as ‘person carrying backpack’ or ‘arm holding knife’.

The technology can also detect unusual patterns, like people moving against the typical flow of foot traffic for a given area (perhaps suggesting fleeing from something/someone) or rapid crowd gatherings that could indicate that an incident is taking place.    

Beyond identifying immediate risks, AI has powerful investigative capabilities.

It can quickly review vast amounts of data that would take a person much longer.  

If a child goes missing, reviewing footage from tens or hundreds of cameras would take significant man-hours.

When a child’s safety is on the line, this is too long.

In contrast, AI can rapidly scan footage and apply filters, like height, clothing type, or clothing colour, to quickly identify and track potential candidates.   

AI can highlight trends and identify potential causes.

For example, installing a handrail could reduce the number of slips occurring in one place.

Even more insights emerge when analysing data from a large area over a specific period.  

Take video analytics of traffic patterns; it can identify hot spots for congestion or even accident-prone roads. This can then inform decisions about new signage, layouts or speed limits.   

Is AI-driven surveillance the future?   

As a tool for our public space surveillance teams, AI offers too many opportunities for improved risk detection and faster incident investigation and can play a critical role in protecting our towns and cities.  

Undoubtedly, AI also introduces challenges of its own.

However, the truth is that the most successful deployments of AI for public space surveillance are likely to be by organisations that are realistic about these challenges and by authorities who acknowledge public concerns and take time to understand how they can be addressed using available solutions. 

By making sure the technology is used appropriately, proportionately and ethically, authorities can confidently adopt AI in a way that benefits all.   

This article was originally published in the October Edition of Security Journal UK. To read your FREE digital edition, click here.