Facial Recognition in Spotlight as Police Arrest UnitedHealthcare Shooting Suspect

With news breaking today about the arrest of a suspect in the killing of UnitedHealthcare CEO Brian Thompson last week, the potential utility of facial recognition technology for police work is getting a larger share in the media spotlight.

Thompson was shot in broad daylight on the streets of New York City by an unknown assailant on December 4th. Later that same day police released screenshots from security cameras depicting the suspect’s partially obscured face.

On Friday, Donnie Scott, CEO of IDEMIA North America, offered some insights in an interview with CNN. Scott’s expertise shed light on the capabilities and limitations of this technology in criminal investigations.

Scott began by addressing the quality of the suspect’s images, stating that they were “very good” from a quality perspective for facial recognition purposes. He explained that police “would say that odds are if that image were in the database, we would match against it.” This highlighted the potential effectiveness of facial recognition technology when provided with clear, high-quality images.

However, Scott emphasized that the success of facial recognition depended on whether the suspect’s image existed in law enforcement databases. He pointed out that, contrary to popular belief, most Americans are not in these databases.

Scott elaborated on the challenges of accessing various image repositories, such as DMV databases, explaining, “The state of New York does not have access to the DMV database for law enforcement purposes by law. And so it requires cooperation, information sharing, and a reason and willingness by the respective agencies to be allowed to share that by law.”

Regarding the comparison between human recognition and facial recognition software, Scott acknowledged that human experts could sometimes be more effective, especially with partial face images. He explained that facial recognition technology’s primary role was to assist law enforcement by narrowing down potential matches.

“Our job is really to make law enforcement’s job easier,” he said. “If you think about the real basic use of facial recognition technologies, it’s to sift through the millions and millions of images to get it down to a small, small set that the expert can use to make the identification.”

Scott also highlighted another crucial application of facial recognition technology in this case: tracking the suspect across multiple video streams. He stated, “We may not know who the perpetrator is, but we can use the same set of technologies to say that perpetrator has been in these seven video streams doing these activities across this series of time.” This capability can significantly speed up the investigation process, as Scott noted, “What used to take us days and weeks is now in minutes and hours.”

All that having been said, the suspect taken into custody today appears to have been apprehended as a result of a tip from a sharp-eyed McDonald’s employee. But that still leaves room for the use of facial recognition to match him to images previously captured through surveillance cameras. At the time of writing, the New York Police Department is keeping tight-lipped about any such developments.

Sources: YouTube, CNN

–

December 9, 2024 – by Cass Kennedy and Alex Perala