Poor Quality Images Hamper Scottish Police Facial Recognition Efforts

Scottish police are facing a major roadblock in their use of facial recognition technology. A new report reveals that only 2 percent of biometric searches return a match, largely due to poor-quality custody images stored in national databases. The findings have sparked fresh calls for policy changes and technological improvements.

Biometric Matching Hit by Low-Quality Images

A joint review by Scottish Biometrics Commissioner Dr. Brian Plastow and His Majesty’s Inspectorate of Constabulary in Scotland (HMICS) has flagged a serious problem: many custody images collected between 2019 and 2024 are unusable for forensic facial recognition. This means police are struggling to identify suspects effectively.

Police Scotland has acknowledged the issue and is reviewing four recommendations made in the report. These include:

  • Developing a specialized policy for biometric searches, addressing legal, ethical, and retention concerns.
  • Enhancing image resolution to prepare for integration with the UK Home Office Strategic Facial Matching Project.
  • Evaluating the need for additional training for officers handling biometric searches.
  • Improving data management and analysis to assess the effectiveness of current practices.

police facial recognition Scotland

A System Underperforming

Right now, facial recognition technology isn’t delivering much value to Scottish police. The numbers tell the story: just 2 percent of searches yield a potential match. In the 2023-24 period, police attempted facial recognition searches 4,006 times, with disappointing results.

To put this into perspective, other UK police forces rely heavily on biometric searches. A review of Police National Database (PND) usage shows:

  • The Metropolitan Police conducted over 31,000 searches in 2023-24.
  • Greater Manchester Police ran 5,290 searches.
  • South Wales Police made 1,727 searches.

These figures highlight a stark contrast in how effectively different forces are using facial recognition.

The Role of Cognitec’s Algorithm

Police Scotland’s biometric system is powered by Cognitec’s FaceVACS-DBScan ID 5.5.0 algorithm, which was upgraded in November 2021. The report includes a review of the algorithm’s performance in National Institute of Standards and Technology (NIST) testing, emphasizing the importance of understanding potential biases in matching individuals from different demographic groups.

Scottish authorities are being urged to examine how the algorithm operates, ensuring it meets accuracy and fairness standards before expanding its use.

Next Steps and Future Challenges

Police Scotland has started addressing the report’s recommendations. A comprehensive biometrics policy document is expected by October 31, 2025. This policy is set to enhance oversight and governance for officers using the PND.

Looking ahead, a major question remains: should Police Scotland join the Strategic Facial Matching Project under the UK Home Office Biometrics Programme? Doing so could provide access to live and operator-initiated facial recognition, along with an improved algorithm for retrospective searches. However, the report warns that significant hurdles must be cleared before adopting any new biometric technology.

The findings put a spotlight on the broader debate around facial recognition’s role in law enforcement. With privacy concerns and ethical considerations at the forefront, the pressure is on Scottish authorities to get it right.

By Dayna Bass

Dayna Bass is a talented news writer at our website, delivering compelling and timely stories to our readers. With a passion for journalism and a keen eye for detail, Dayna covers a wide range of topics, ensuring that our audience stays informed about the latest news and developments. Whether it's breaking news, investigative reports, or human interest stories, Dayna's articles are meticulously researched and written with clarity and accuracy.

Leave a Reply

Your email address will not be published. Required fields are marked *

Related Posts