South Wales Police facial recognition system accuracy improving but in-the-wild results lag

Improvements to the facial recognition system used by South Wales Police has raised the technology’s effectiveness to confirmed match rates of 26 and 22 percent, though a small-scale field trial indicates it has the potential to reach 76 percent accuracy, according to a report by Cardiff University’s Crime and Security Research Institute.

The test found that the technology used by South Wales Police, which includes NEC’s NeoFace Watch, struggles to match people in crowded environments, when they are wearing glasses or hats, or when moving quickly, according to the report, titled “An Evaluation of South Wales Police’s Use of Automated Facial Recognition.” Judging the system’s results with a new M20 algorithm introduced in late 2017, the report found that the “locate” function, which matches live video feeds against a watchlist, delivered 50 percent confirmed false positive matches. The system would also freeze or lag when attempting to identify several faces in the same frame. For the “identify” function, which matches still images against a police custody database, operators rejected 9 percent of the matches identified by the system, but 60 percent of images were rejected by the system for being of too poor quality. The camera equipment does not function well in low light, according to the report.

The researchers conclude that at this time, the system should be considered “assisted” facial recognition, rather than “automated” facial recognition.

A total of 18 arrests were made in live situations, with the assistance of the “locate” capability, while more than 100 charges were filed in a span of eight to nine months as part of the AFR Identify operation, which ended in March of this year.

“There is increasing public and political awareness of the pressures that the police are under to try and prevent and solve crime,” says Professor Martin Innes, who led the evaluation. “Technologies such as Automated Facial Recognition are being proposed as having an important role to play in these efforts. What we have tried to do with this research is provide an evidence-based and balanced account of the benefits, costs and challenges associated with integrating AFR into day-to-day policing.”

The South Wales Police force is under sever cost pressures, according to the Financial Times.

The force has faced criticism for its use of the technology, after disappointing early returns, but touted a dramatic improvement from the new algorithm, with a mere ten false matches out of more than 44,000 face scans at music festival the Biggest Weekend in Swansea in May.

“The report provides a balanced perspective of our use of the technology and hopefully it will help to demystify some of the misunderstandings and misinformation that have proliferated across the press,” comments South Wales Police Deputy Chief Constable Richard Lewis.

“South Wales Police remains committed to the continuous use of the technology in a proportionate and lawful way to protect the public, whilst also remaining open and transparent about how and when we use it.”

“As highlighted [in the report], tech should be seen as enhancing human capabilities not replacing them,” Director of Research at think tank Reform Eleanora Harwich told the Financial Times. “Developing a strong evidence base around the effectiveness of these types of tools is crucial.”

Article Topics

accuracy  |  biometrics  |  facial recognition  |  police  |  UK  |  Wales