Webinar presenters Mark Dolfi, Stephen Bevan, Tim Meyerhoff, Michael Cuti, and Jason Thompson answered a number of your questions after their presentation, “Iris Recognition in Law Enforcement.” Here are just a few of their responses.
Audience Question: So, for the panelists from your participating agencies today, so, Steve and Jason, what do you know now about implementing and writing an Iris ID program that you wish you would have known when you first started out?
Jason Thompson: This is Jason. I guess for it was that it was a lot easier to implement than we thought. We really did a slow rollout of the iris technology. The Border Patrol is made up of about 20 sectors and more or less about 135 Border Patrol stations. We started our rollout across all 135 Stations in 2014 and we just wrapped it up this year. I think that you know, if I were to realize that people would have picked it up as easy as they have and that the benefits are what they are, we would have pushed for that to be faster.
Mark Dolfi: Steve stepped out but I’m here and one of the things that when we started a few years ago. As Jason said it was a lot simpler to get started than we expected and looking back on it now, I think that we could have gone back and added more facilities at that time because with the higher number of overrides by our personnel or the uncooperative inmates. We weren’t collected as much as we thought we could but, I think that when we expand to new facilities, this go around, we’ll be able to capture a lot more and get everybody trained on how to properly use the camera itself.
Audience Question: What’s the probability that two or more people could have the same iris identifiers?
Tim Meyerhoff: I’ll take that one. The standard algorithm, that we release, like a commercial algorithm that’s used for border control, national ID, like in Mexico or even the access control system, the probability of a false accept on a single eye meaning that’s a pirate, a pirate recognition we call that. The threshold is set at one in one point two million. So, when we do both eyes of capture at the same time, the probability of a false accept is one in a few billion. So, it’s extremely accurate. The benefit as compared to fingerprint or face is the scalability. It’s very cost-effective to scale it computing power-wise.
Audience Question: What happens if the eye becomes damaged? Is there accommodation there? Do you also take fingerprints in conjunction with the iris? How do you address that situation?
Tim Meyerhoff: Multimodal is the way to go, like Jason mentioned. Border Patrol and many of these other agencies like LACRS they do multimodal enrollment. Face, finger, and iris are enrolled. There are only a few situations where an iris would become damaged. Actually, we’ve done a lot of studies with DHS with respect to, for example, ophthalmologists which do cataract surgery. The iris is very stable, the enrollment is still OK after that. Just like Jason has mentioned with fingerprints, they can become damaged, but the probability is a lot lower for iris.
Jason Thompson: What we do is when we capture, If somebody has an eye that’s missing, it’s been amputated, you can label it as such. Then we’ve also had issues where maybe they’ve gone to the hospital, and the hospital has put an eye patch on for an eye injury, and then, we just label it as unable to capture. That way the database knows that on this day, we were only able to capture one iris but on, on a future date, they might be able to collect two again. Whereas, if you label at his amputated or missing. Hopefully, they will never – well, I would anticipate they were not going to grow an eye back
Audience Question: So also, I’m thinking of people with an artificial eye like a glass eye that you would also label it that way. It’s just unable to take it or Tim what would be the right.
Tim Meyerhoff: And so, our camera technology includes what we call counter-measure. So, it’ll detect the contact lens and it will also detect prosthetic eyes. The camera actually can detect that when it goes to get an image, if there’s a contact lens or glass eye, prosthetic eye, the camera technology has that. That’s one of the cornerstones of our business actually is what we call fake eye detection or prosthetic eye contact lenses detection. Those are very important facets of the business.
Audience Question: Tim, could you do me a favor and go back to the contact page? I’ve got several people asking for you to go back to the page. with everybody’s e-mails, they can jot down. There we go.
Tim Meyerhoff: Yeah, this is also available as the handout. I think everybody will get a link to the slide deck, right?
Audience Question: Has iris been used in human trafficking or sex trafficking? Will this be a viable application of the technology? And, if so, are there any benefits or anything that we need to think about in terms of specific human trafficking or sex trafficking?
Tim Meyerhoff: I am not directly aware of databases doing that or are doing that most of the work that has been done by NATO and the National Institutes of Health has to do with refugee management. There’s a very lot of activity in those particular areas, but specifically for sex trafficking, you know, I’m not aware of that in particular, from a domestic perspective but we have global deployments in Middle, East, Africa, you know, all over the world. So, I don’t have a direct answer to that, but I know we do a lot with health care and refugee management, in particular, you know, for benefits management, for a positive ID. A lot of those individuals like Jason that mentioned a lot of these people are laborers, their fingerprints are damaged. So, to do accurate ID with fingerprint, or even face is very challenging. With Iris very easy to do.
Stephen Bevan: I’d like to add to that. To get up to get a match on a person that you know, whether they’re sex trafficking or you know, human trafficking, they have to be registered. So, if we don’t have a database with those eyes in there now, then, it doesn’t matter if we have a great image from a victim that we were able to capture their irises that may be a train station or an airport but because there’s no database to search again. So, they have to be enrolled and there’s been talk in LA County with LACRS that we’re looking for a registration typed self-registration, but we have a long way to go with that. So hopefully we can get that But that’s different because we’re using the criminal databases only, so that doesn’t work for us but as a person out in the public, they would have to be able to give those eyes kind of like what they do at the airport’s where they self-register their eyes and then there’s a whole other public scrutiny and policies to follow for that.
Audience Question: How does drug use or medications or alcohol affect the scan of the iris for identification purposes?
Tim Meyerhoff: There is very little impact with respect to the dilation of the pupil. It’s directly related to that and my algorithm guys can work they could address that better than I can, but it’s referred to as a rubber sheet method. That being said, if the pupil is extremely dilated the software says, hey, wait a minute, there’s something going on here. So, if it’s like a saucer and there’s no iris exposed or the amount of iris compared to the pupil that sets a flag in the software that says, hey wait a minute something may be wrong with this coming in. So that’s part of the whole quality assessment, a back-end piece that we do like post-capture. First, we check for liveness and contact lenses, and then we can drill down into areas like that with respect to pupil dilation. That is not a direct indicator of any kind of drug use, but it is a measure that is available.
Click Here to Watch a Recording of “Iris Recognition in Law Enforcement.”