What is live facial recognition and why are Police Scotland seeking your views?

Police Scotland is seeking the views of the public on the use of live facial recognition (LFR).

The force, the Scottish Police Authority and the Scottish Biometric Commissioner are seeking opinion on the use of LFR via an online survey.

Officers say it isn’t a consultation but an ongoing public conversation about the potential use of the technology.

It is already in use in the private sector, particularly by retailers, but there is currently no piece of legislation for the use of LFR in Scotland or the UK as a whole.

What is live facial recognition and how accurate is it?

Facial Recognition technology uses Artificial Intelligence (AI) to identify individuals in a digital image such as a video or photo.

It relies upon an algorithm that is trained to detect what a face is. The algorithm will then be trained to recognise faces and distinguish one facial image from another.

It does so by measuring distinguishable features of an individual’s face and assessing real-time, live video footage of crowds who are passing a camera.

Facial Recognition technology uses Artificial Intelligence (AI) to identify individuals in a digital image. iStock

It automatically compares the images of those in the crowd against a database of people on a predetermined bank of images, known as a watchlist.

A review of LFR usage by the Metropolitan Police Service (MPS) found that 500 hours of surveillance footage were captured between January and June 2024.

Further analysis of the 79 deployments conducted in this timeframe found that police stopped one person every 55 minutes, with an arrest every 128 minutes.

The review, conducted by Stopwatch, concluded that there was limited evidence of the efficacy of LFR.

However, Big Brother Watch published findings raising concerns around the accuracy of LFR.

Published in 2018, this report found that the use of LFR by the Metropolitan Police Service had a 98% accuracy rate and that South Wales Police (SWP) use had a 91% accuracy rate.

However, the Biometrics Institute concluded that there was insufficient data in the report to assess the accuracy of LFR.