Celebrating Spartans: Facial Recognition


EAST LANSING, MI (WLNS) – A Spartan created a software program that’s helping law enforcement fight crime around the country.

He’s a distinguished professor of computer science and engineering at Michigan State University and he’s considered by many to be the “Michael Jordan” of his profession.

Dr. Anil Jain created a powerful new facial recognition tool that locally is helping the Michigan State Police solve more crimes making our communities safer.

Thirty year ago, investigators would have to look through mug shot books to identify a suspect. Now with the help of a computer algorithm that Dr. Anil Jain created law enforcement agencies are able to search a database of 30 million images in a matter of seconds, going way beyond what the human brain is capable of.

“We can recall only a very small number of faces, 1,000, 2,000. But in law enforcement applications we have to compare a photo of a suspect against tens of millions of photos,” said Dr. Anil Jain, distinguished professor of computer science, Michigan State University.

And that is paying off for law enforcement agencies nationwide, especially for the Michigan State Police.

“The quicker that they’re able to apprehend these people, the less victims of crime there are going to be.”

Pete Langenfeld, the manager of the Michigan State Police’s Digital Analysis and Identification

Section says when it comes to identifying a potential suspect; Dr. Jain’s algorithm has given his department a powerful tool to help attach a name to a face.

“Within the last 5 years, the technology has improved greatly, giving MSP and law enforcement the ability to identify people who try to remain anonymous, say on social networking and things like that, not using their name, that we have the ability to identify them a lot quicker,” said Pete Langenfeld, MSP digital analysis and identification section.

Langenfeld says combining that with great detective work out in the field, his department is having a much greater degree of success.

“This unit here, with the two analysts we have doing this probably sending out two or three IDs, potential leads, investigative leads per day. And that’s for homicide, rape, financial fraud, credit card fraud, cloned credit cards.”

Crime stats show that more and more people are becoming victims of credit card and identification fraud. Facial recognition is helping to identify the criminals who wreak havoc on people’s lives by stealing their personal information.

“They go into these retail establishment and use a card. So then we’ll work with that image that is taken from there and we’ve been able to ID quite a few people just from those images.”

Let’s say an image was taken from a surveillance video with the person looking off to the side. MSP can create a 3D image of that picture and then rotate it to the front so you can see the person’s eyes.

“Eyes play a very important role in facial recognition because we use the center point of the two eyes to do the alignment. So when you compare the two faces, they have to be in the registration. And we typically use the eye positions to align them. But it the eyes are not visible the alignment becomes difficult,” said Dr. Jain.

The image is put into the Statewide Network of Agency Photos or SNAP system. Then, by using computer program from a leading biometrics company that licensed Dr. Jain’s algorithm, that image is run against tens of millions of images in the database and within minutes it produces an investigative lead for detectives.

“Using the algorithm that Dr. Jain has helped developed is night and day,” said Langenfeld.

So we asked Dr. Jain to test his facial recognition algorithm using a picture of Sheri Jones, one that was taken outside a couple of years ago. This time it was run through his data base of 120 million images.

After just a few minutes the computer program produced the top five results and there clear as day were two of Sheri Jones’ professional pictures, a 100 percent match.

6 News decided to try another scenario with the state police. Let’s say a state trooper pulls someone over, but the driver can’t produce any identification. The trooper could then use his or her phone to take a picture of the driver and then send the image to be identified using MSP’s mobile facial recognition technology. MSP analyst Angie Yankowski tried it with Sheri Jones.

“This process of mobile facial recognition searches takes approximately two minutes or less and it is searching the photo I just took of you against approximately 40 million images,” said Angie Yankowski, analyst, Michigan State Police.

Again in just a matter of minutes 10 images were returned from a search through MSP’s SNAP system and three of them were Sheri Jones. Again a 100 percent match.

Reporter: “Did you ever imagine having this type of technology at your fingertips?”

Angie Yankowski: “Never. It’s speeding up the process for them, if you will. Versus having to take them back to jail to try to identify who they are. Now they have this information a lot quicker, potentially on the side of the road.”

Copyright 2021 Nexstar Media Inc. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.

Michigan Headlines

More Michigan