Consumer and civil rights groups are questioning the results of a facial recognition trial conducted by New Zealand’s supermarket chain Foodstuffs. The retailer has been deploying the technology in its stores to prevent retail crime.
One of the main concerns about the technology is the accuracy of facial recognition on minority populations with darker skin. According to independent consumer protection organization Consumer NZ, the Foodstuffs trial did not gather information on the ethnicity of individuals misidentified by facial recognition. The retailer also did not provide information on how it reported statistics on mistaken identity, The New Zealand Herald reports.
Foodstuff says that its approach to deciding on matches was consistent with a study from the U.S. Commission on Civil Rights. When the system detects a match, two specially trained team members still must agree that a match was made before following up with an intervention.
“The software we’re using was designed and developed in Australia, by an Australian organization. They use multiple datasets from around the world including locally acquired data from Polynesian, Indigenous Australian, Asian, African American and American groups,” Foodstuffs says.
Foodstuff’s trial was designed and independently tested by evaluation and research firm Scarlatti and concluded in September. The results are currently being scrutinized by the country’s Privacy Commissioner Michael Webster who plans to announce its results by the end of 2024.
New Zealand’s debate on facial recognition will also impact other firms, including crime intelligence platform Auror which has recently raised US$48.7 million to fund its global expansion.
Meanwhile, Foodstuff is facing criticisms from legal experts and groups such as the New Zealand Council for Civil Liberties over other issues, including the lack of informed consent, data privacy and hacking risks.
The company says its system automatically and immediately deletes images of shoppers unless their images match the store’s record of “offenders and accomplices.” The firm also addressed the issue of minors noting that no images of minors are being entered into any facial recognition system.
“During the trial, the store datasets of enrolled offenders were reviewed as a part of the trial evaluation and no minors were identified,” says the firm’s spokesperson.
The retailer has announced that it will continue using facial recognition in the same 25 stores that first trialed the technology using the same privacy protocols and processes.
Article Topics
biometrics | criminal ID | data privacy | facial recognition | New Zealand | retail biometrics | video surveillance
