UK Home Office live facial recognition adoption begins with POC at ports

The UK government is joining its police in embracing facial recognition with the Home Office and Immigration Enforcement planning a proof of concept at ports for immigration enforcement. The POC is revealed in a trio of policy papers on how live facial recognition should be operated, how immigration enforcement can use face biometrics, and the equality impact of LFR.

Immigration Enforcement, Border Force and police partners will begin the time-limited POC by the end of 2025 to detect “persons returning in breach of a Deportation Order (DO) for law enforcement purposes.”

The decision comes in the context of a steady increase in the use of live facial recognition by police in the UK and the proposed introduction of a national digital ID with the explicit purpose of reducing illegal immigration.

All three papers are dated October of this year, but were published online on Wednesday.

Home Officer’s LFR SOP

A 25-page paper sets out “Live facial recognition standard operating procedures” sets out the authority under which the technology can be deployed, considerations for deployment decisions and watchlist generation and stipulates three mandatory assessments that must be carried out. It defines roles for LFR operations and describes required evaluations that must be completed post-deployment. The paper also establishes rules for data retention and management. At least some of these, such as the immediate and automatic deletion of biometric data that does not generate an alert, match those implemented by UK police according to policies set by individual forces like the Met.

The SOP includes ensuring that the minimum biometric facial matching threshold is above 0.64, in accordance with NPL recommendations, notifying the public of the time and duration of deployments ahead of time and taking appropriate privacy expectations into account when determining these. A person avoiding “the Zone of Recognition” does not justify enforcement alone, the paper states.

For now, the watchlist can only be populated with images of people subject to a deportation order.

Home Office’s live facial recognition system should be able to deliver a “false alert rate” below 1 in 1,000.

A data protection impact assessment, equality impact assessment and self-assessment based on the Surveillance Camera Commissioner’s code of practice must be reviewed, amended or adopted for each deployment, as necessary.

Immigration Enforcement policy

“Immigration Enforcement facial recognition policy” is explained in a 45-page document. This paper reviews facial recognition technology and how law enforcement uses it, and covers the intentions, processes and assessments involved in its application for immigration enforcement.

The document is subject to change based on guidance from the Information Commissioner and the Biometrics and Surveillance Camera Commissioner, as well as in legislation or biometric technology. Immigration Enforcement places its context within the various policies around immigration and facial recognition, and sets out plans for documentation to explain the system to the public.

Though it is an Immigration Enforcement POC, it will be carried out with staff and equipment from South Wales Police and Greater Manchester Police.

The policy reviews the technical requirements for cameras, and suggests “The use of an `attractor’ to direct a subject’s gaze towards the camera may help to obtain better quality images.”

“True recognition rate,” false alert rate (false positive identification rate) and “recognition time” are defined as key performance metrics, but other than the 0.1 percent for FAR targets are not specified.

Accuracy, bias and disproportionality are dealt with at length, including with a review of the NPL study.

The 16-page “Live facial recognition: equality impact assessment (EIA)” emphasizes the importance of using the 0.64 match threshold and identifies steps taken to mitigate indirect discrimination against people based on a broad range of characteristics.

Article Topics

biometrics  |  border security  |  facial recognition  |  immigration  |  law enforcement  |  live facial recognition  |  real-time biometrics  |  UK