ARTICLE

Iowa DOT using facial recognition technology

Use not widespread in Eastern Iowa, but technology is improving

A method of facial mapping is demonstrated in Clearwater, Fla., on June 4, 2013, by Scott McCallum, systems analyst and co-administrator of the facial recognition program for the Pinellas County Sheriff's Office. The Pinellas County Sheriff's Office utilizes one of the most advanced facial recognition programs available to law enforcement in the country. (Photo by Edward Linsmier for The Washington Post)
A method of facial mapping is demonstrated in Clearwater, Fla., on June 4, 2013, by Scott McCallum, systems analyst and co-administrator of the facial recognition program for the Pinellas County Sheriff's Office. The Pinellas County Sheriff's Office utilizes one of the most advanced facial recognition programs available to law enforcement in the country. (Photo by Edward Linsmier for The Washington Post)

Even though Iowa investigators have access to facial recognition technology, those in the eastern part of the state say they’ve rarely made much use of it.

Iowa is one of 26 states that allow law enforcement agencies to search — or request searches — of state ID photo databases in an attempt to learn the identities of people considered relevant to investigations, according to the Washington Post.

The Iowa Department of Transportation first began using image verification software in 2008 with the intention of preventing driver’s license fraud and identity theft. Since that time, it also has been available to law enforcement agencies — state, local and federal — who might be trying to identify an unknown suspect or witness in a crime.

Mark Lowe, director of the DOT Motor Vehicle Division, said police do not have direct access to the software, but can submit a photo for image verification after proving legal or criminal predicate, which means it will be used for law enforcement purposes. They also fill out a Driver Privacy Protection Act agreement, which prevents police from releasing and using certain personal information from state motor vehicle records.

The 12 million photos in that database represent at least 2.1 million people in the state. Every person with an Iowa driver’s license or state ID is included in the data.

How it works

Once police prove legal or criminal predicate, the image they wish to compare to the database is scanned and turned into a grayscale image.

Next, the facial recognition program locates the eyes and measures the distance between them and key points on the face. That data is stored as a template in a database and can be compared against existing and future measurements from other photographs.

If law enforcement requests to have a photo scanned, the image goes through the same process and is compared against other stored measurements with hopes of finding a match.

The system doesn’t take gender, age or race into account, which can help in situations where a person might try to change their appearance. The software can be adjusted to show varying degrees of precision in a query.

Once the system matches the image with templates, staff weeds out obvious mismatches before trying to identify the subject. Police do not rely solely on a match in making an arrest.

“You would still need to go back and do legwork to coordinate that event on that day,” said Paul Steier, a major with the Motor Vehicle Enforcement Office. “It’s not the probable cause we need to make an arrest or get a search warrant.”

The Iowa Department of Public Safety is in the process of creating a similar database for sex offenders. Steier said the state hopes to eventually digitize all mug shots so they can be entered into a facial recognition database to expedite the identification process.

In Eastern Iowa

Steier said his office gets about six requests from law enforcement a week, and, so far, they’ve helped police develop leads that resulted in the capture of fugitives and identification of child predators.

Though Steier said the software also has helped identify at least 700 cases of identity theft since it was initiated, Eastern Iowa law enforcement officials said they haven’t made much use of the technology.

Col. John Stuelke, with the Linn County Sheriff’s Office, said photos of unidentified subjects often come from surveillance cameras which tend to be grainy, black and white, and hard for the software to read.

“Obviously we’d like to have success, but we haven’t had a good photo,” Stuelke said, adding the department has requested verification a handful of times since it was available. “As things get better with technology, I’m sure we’ll be able to make it work better for us, but at this point it really doesn’t.”

Johnson County Sheriff Lonny Pulkrabek said his investigators have not made requests to use the image verification software.

Privacy concerns

Because the technology is relatively new, Ben Stone, executive director of the Iowa branch of the American Civil Liberties Union, said the DOT should set up more specific regulations for how the software is accessed and used by police.

“You don’t want police to be able to go in there and look around without investigating something specific that their supervisors have authorized,” Stone said. “You also don’t want a situation where you get too many people identified as potentially matching and those people end up getting forced to interact with police officers and they haven’t done anything wrong.”

Lowe said he felt the DOT’s checks and balances were sufficient, adding the department tracks who is accessing the database and why it’s being accessed to prevent abuse.

Though 11 states with facial recognition systems do not allow law enforcement to access them, the Washington Post reports that the general trend is moving toward more sophisticated software with more access.

Lowe said the state is currently in the process of upgrading its image verification software, which will improve its accuracy.

Give us feedback

We value your trust and work hard to provide fair, accurate coverage. If you have found an error or omission in our reporting, tell us here.

Or if you have a story idea we should look into? Tell us here.