New NEC algorithm dramatically improves accuracy of South Wales Police facial recognition
South Wales Police have deployed a new algorithm to make their automated facial recognition (AFR) system much more accurate, making only ten false matches out of 44.468 face scanned during the Biggest Weekend event in Swansea, WalesOnline reports after receiving data through a Freedom of Information request.
Further, two alerts were issued for matches confirmed by the operator, with one person arrested for an outstanding warrant. The match was against one or more of three watch lists created by police specifically for the event. One list was based on suspected organized crime groups associated with music festivals, one was based on individuals wanted by South Wales Police, and one for wanted suspects from the specific area.
“We have been provided with a new algorithm and application by our supplier, NEC, which has 12 months more research and development from the project’s initiation,” South Wales Police said in a statement. “With each deployment of the technology we have gained confidence in the technology and this has enabled the developers at NEC to integrate our findings into their technology updates.”
The police statement also said 16 individuals had been arrested after being identified with the AFR Locate function between May 31 and December 22, 2017, and 13 more were arrested from February 3 to June 24, 2018. AFR Locate is the force’s real-time CCTV The AFR Identify capability, which enables police to compare images of persons of interest with half a million custody images, has led to 450 arrests since July 2017.
The new and improved results are a dramatic change from the application of the AFR system at a 2017 Champion’s League match, at which 92 percent of matches were false positives. The ineffectiveness of the system contributed to criticism and a pending legal challenge brought by former Cardiff councilor Ed Bridges.