Toronto Police talk responsible use of NEC facial recognition in first case to identify murderers

The initial use of the Toronto Police Service’s facial recognition is examined in some depth by The Toronto Star, which reports that the case shows both the technology’s impressive power and the need for effective policies and processes that recognize its limitations.

Toronto Police contracted the technology from NEC in 2018 for $450,000, according to the Star, for use matching suspects against its internal database of 1.2 million mug shots.

The lead investigator in a targeted murder in the city in March of 2018, Jason Shankaran, concluded that the victim did not know his killers, and they would have to be identified through their path to the scene of the crime. Police began collecting security video footage, and found the two suspects were wearing distinctive clothing. A get-away vehicle was also identified, but it was stolen.

More surveillance camera footage was used to track the vehicle to an approximate location, and it was then found by a detective. Based on where it was found, further video footage was obtained, showing the suspects lived in the area, and eventually their faces.

In the past, they may have circulated photos, but that could also tip the suspects off to the police being on their trail, and evidence may disappear before their capture. Instead, the clip was sent to members of the service’s facial recognition team, who have received training on the technology from the FBI.

It returned 200 potential matches, which were narrowed down to three “potential candidates,” two of whom were eventually charged, while the other turned out to be a false positive.

“We needed to treat this information as an investigative aid,” Shankaran notes. “It’s not a fingerprint or DNA. There’s still a need to corroborate the potential candidate as we would any other information gleaned from a witness, or otherwise.”

“In the end, the decision made by myself or a member of my team, although based on our training and experience, is totally subjective,” says Detective Sergeant Shawn Meaney of TPS’ facial recognition team. “This is why any information provided to an investigator is an investigative tool only.”

Shankaran is convinced of the technology’s value, as long as it is used properly.

“Any investigator must use it cautiously, and corroborate any selection and use it to include or exclude people,” he says. “It’s a powerful tool, and it’s not just for accused, it could also locate a witness.”

When it came to the trial, a conviction was secured, and the biometric evidence was not challenged. “I think we did a good job at articulating the limits of where it could take us,” Shankaran told the Star.

Toronto Police have also tested Clearview AI. The tests were “informally” carried out by some officers, who used it more than any other service in the country, the Star writes.

The Australian Federal Police initially denied it had used the company’s software, but has since acknowledged that officers trialed it in late-2019, according to the ABC.

Seven members of the Australian Centre to Counter Child Exploitation (ACCCE) registered for trials and performed searches between early November 2019 and late January 2020. The revelation has prompted questions from parliamentarians and rights advocates about how the AFP could have been unaware of technology being used by its officers.

Using its technology from NEC, TPS conducted 1,516 searches with facial recognition between March and December of 2018, with potential candidates returned for about 60 percent, according to an internal report to the police oversight board. Of those, 80 percent led to criminals being identified.

Article Topics

biometrics  |  Canada  |  criminal ID  |  facial recognition  |  forensics  |  NEC  |  police  |  Toronto police