Black woman matched by facial recognition alleges police misconduct in lawsuit

Another person in the United States is alleging a wrongful arrest based on a false match by facial recognition software used by police. It is the sixth allegation of facial recognition prompting the arrest of the wrong person in America, and the third in Detroit. All six incorrectly identified people are Black.

A lawsuit has been filed by Detroit resident Porcha Woodruff, who was 8 months pregnant when she was arrested for carjacking and robbery, The Detroit News reports. A photo of the suspect was matched with an eight-year old mugshot photo of Woodruff, despite the availability of her recent driver’s license photo, according to court filings.

Detroit police extended a contract for facial recognition software from Dataworks Plus and Rank One Computing in 2020.

Woodruff’s attorneys write that the officer who obtained the warrant for her arrest neglected to mention the age of the reference image used for the biometric match. If that information had been disclosed, they say, the warrant may have been denied for lack of probably cause.

When Woodruff’s image was returned as a match candidate for the probe, her eight-year old mugshot was placed in a photo lineup, where it was picked out by the victim.

The case against Woodruff was dismissed during a preliminary examination for lack of evidence.

When she was arrested, Woodruff was held for 11 hours, and she was treated for dehydration on release, she told the New York Times, adding further emotional weight to the story. The person who participated in the carjacking, and whose image was captured by a security camera, was not visibly pregnant.

Detroit Police Chief James White said the allegations in the lawsuit are “very concerning,” and the department will perform “additional investigation,” as quoted by The Detroit News. He also told the police oversight board that a supervisor had reviewed the case before authorizing the warrant affidavit, and proper procedure was followed.

The ACLU filed a federal lawsuit seeking to make Detroit police stop using facial recognition in 2021, in response to the false arrest of Robert Williams. The arrests of Williams, Woodruff, and Michael Oliver in Detroit all took place within the last three years. In Williams’ case, White’s predecessor James Craig told 60 Minutes that the arrest resulted from “sloppy, sloppy investigative work.”

Psychology professor Gary Wells, who has researched the reliability of eyewitnesses, says that facial recognition systems with enough reference images will always find one that looks like the probe, and the eyewitness will usually assume that the correct person’s image is among those they are shown. This makes the combination unreliable, he says.

Assessments by the National Institute of Standards and Technology show that many leading facial recognition algorithms are just as accurate for Black men as other groups of people, but others are not. It also shows that even measuring demographic fairness in facial recognition algorithms, let alone whole systems, remains a challenge.

Article Topics

accuracy  |  biometric matching  |  biometric-bias  |  biometrics  |  facial recognition  |  false arrest  |  police  |  United States