facial analysis Archives

Does Your Consultant Offer Biometric Domain Fluency?

Considering the right consultant for your biometric firm often comes down to a single, make-or-break dynamic: biometric domain fluency.

But that’s just a fancy phrase for knowing what you’re talking about.

“The Longest Hour.” Google Lyria. Public Domain.

Consider what happens when a product marketing consultant at a biometric firm does NOT possess biometric domain fluency.

Off-target content

Imagine that you’re writing your first blog post for a biometric company. So you figure you’ll get educational by telling your prospects how “Gender Shades” identified huge accuracy problems, but your firm’s facial recognition algorithm has overcome them.

Consultants with biometric domain fluency know the difference between facial recognition and facial analysis. — Google Gemini.

Congratulations, novice. You just mixed two distinctly different things up.

I’ve addressed this before, but I’ll address it again. Do you know why that old study was called “Gender Shades”? Because it tested algorithmic ability to recognize gender and race. That is facial ANALYSIS (or categorization), not facial RECOGNITION (or individualization). The National Institute of Standards and Technology (NIST) draws a clear distinction between the two. And if you use a high-quality facial recognition application, the gallery of possible matching candidates may have a white woman next to a black man…because the facial recognition algorithm DOES NOT CARE about gender or race.

And that’s just one example of how someone who doesn’t know biometrics can make a grievous mistake that repels a firm’s prospects.

Losing proposals

Before I launch into this, let me explain that there are a number of reasons why a firm may lose a proposal.

Not submitting your proposal in Times New Roman 12 font with one-inch margins.

Submitting a 16-page proposal when the page limit is 15 pages.

Submitting your proposal Friday at 5:03 pm when it was due at 5 pm.

Here I want to talk about something different: getting the answers wrong.

Let’s say you’re tasked to answer a question about submitting your fingerprints at 500 pixels per inch (ppi). But you want to get some extra evaluation points, so you write this:

“The Super Duper Ridger fingerprint reader EXCEEDS THE REQUIREMENT by capturing fingerprints at 1000 ppi.”

Congratulations, novice. You’re non-compliant.

Granted, latent fingerprint examiners LOVE higher resolution tenprints and latent prints where they can see the finer detail. But for everyone else, high resolution prints can potentially cause huge problems. Longer time to display images on the screen. Longer network transmission time and higher network bandwidth. Oh, and four times as much storage space. If the agency specifies 500 ppi, they mean what they say.

You have to know your biometrics to answer the questions properly.

Flawed analysis

Sometimes, product marketing consultants need to strategically analyze the market. Or the products in the market. Or the competitors who sell the products in the market. Analysis is a minefield in which an inexperienced person can get blown up.

Let’s say you’re asked to identify the leading facial recognition algorithm providers for U.S. intelligence community agencies. Now you, being a smart person who knows the difference between face recognition and face analysis, hone in on NIST’s Face Recognition Vendor Test (FRTE) collection. You go to FRTE 1:N, identify the specific test that best meets the needs of the intelligence community, and sort the list with the most accurate algorithms at the top. You then report results for all the leading vendors…including CloudWalk, Megvii, and SenseTime.

Google Gemini or Imagen; I can’t remember which.

Congratulations, novice. The firm that contracted with you will NEVER get intelligence business.

All three of the firms I mentioned are Chinese firms. Just because NIST or the FBI tests an algorithm doesn’t mean that U.S. agencies will buy it. For legal and practical reasons, the CIA is not going to install a Chinese algorithm on one of its systems.

No point in performing a competitive analysis on firms that will never compete in your markets.

Avoid these problems by contracting biometric domain fluency

You can avoid these problems by contracting with a product marketing consultant who offers biometric domain fluency.

Such as Bredemarket.

My wife passed by me while I was drafting this post, and when I told her I was writing about biometric domain fluency, she was not impressed.

Then again, she is not an identity marketing leader.

But if YOU are an identity marketing leader, consider Bredemarket for your consulting needs.

Watch and book.

Learn about what I can do for your firm and schedule a free meeting.

https://bredemarket.com/watch-book/

Another Great Renaming (From 2024)

In August 2023, I talked about what I called a “great renaming” in which the National Institute of Standards and Technology (NIST) differentiated between its face recognition tests and its face analysis tests: a very important distinction and a critical update.

The ramifications of that renaming persist. Just last week I reminded a biometric firm that its references to “FRVT” were dated.

But what if references to biometrics are dated?

Find…what?

Over twenty years ago a publication called FindBiometrics was established that discussed you-know-what. Fingerprints, faces, irises, and all sorts of stuff.

You would think that a name that incorporated “biometrics” would be all inclusive. It certainly was twenty years ago. But as the industry evolved, the name became a little dated. While biometrics remain critically important, I have to say (with apologies to my former Motorola colleague Edward Chen) that biometrics is not “4” ALL. (You see what I did there.)

The publication realized this also, and performed its own great renaming.

“TORONTO, ONTARIO, CANADA, November 21, 2024 /EINPresswire.com/ — FindBiometrics, a leading news media platform for the biometrics and digital identity industry, is now ID Tech—a refreshed brand identity that reflects the beginning of a new era in the identity technology space.”

Why?

“…the scope of identity tech has expanded to integrate new developments in areas like artificial intelligence, blockchain, and digital ID.”

One example being mobile driver’s licenses, which can utilize biometrics but goes far beyond it. After all, biometrics (something you are) is just one of the six factors of identity verification and authentication. See below.

So now the former FindBiometrics platform is called “ID Tech,” and its URL is now https://idtechwire.com/. And biometrics now shares the stage with other factors.

Biometrics shares the stage. “Revealed” from Google Lyria; Public Domain.

But ID Tech isn’t the only place to learn about identity beyond biometrics. There’s also my book.

Four pages from "Proving Humanity: The Six Factors of Identity Verification and Authentication" by John E. Bredehoft, Bredemarket. Click on the image to purchase.

If You Can’t Make It, Fake It: Generation of Synthetic Faces for Algorithmic Testing

Sometimes it seems like there’s a catch-22 in facial algorithm development. On the one hand, opponents complain: “How do you know these algorithms work if they’ve never been tested on real faces?” Then in the next breath they complain, “You can’t use the faces of real people to test your algorithms! That violates their privacy!”

So what do you do?

Fake it.

There are many ways to create fake faces for enterprise and consumer use, but how do we know that synthetic faces are sufficiently representative of real ones?

That’s the challenges these researchers faced:

“Face recognition models are trained on large-scale datasets, which have privacy and ethical concerns. Lately, the use of synthetic data to complement or replace genuine data for the training of face recognition models has been proposed. While promising results have been obtained, it still remains unclear if generative models can yield diverse enough data for such tasks. In this work, we introduce a new method, inspired by the physical motion of soft particles subjected to stochastic Brownian forces, allowing us to sample identities distributions in a latent space under various constraints. We introduce three complementary algorithms, called Langevin, Dispersion, and DisCo, aimed at generating large synthetic face datasets. With this in hands, we generate several face datasets and benchmark them by training face recognition models, showing that data generated with our method exceeds the performance of previously GAN-based datasets and achieves competitive performance with state-of-the-art diffusion-based synthetic datasets. While diffusion models are shown to memorize training data, we prevent leakage in our new synthetic datasets, paving the way for more responsible synthetic datasets.”

If you want to see the synthetic data these researchers created, and if you have the ability to uncompress tar.gz files (Mac and Windows 11 support this), visit this page.

The May 6, 2026 List of PAD 3 Conforming Solutions

Update to the April 2 version. Added Shufti.

Vendor	Modality	Confirming Lab	Link/Date
Aware	Face	BixeLab	November 2025
BioID	Face	Tüvit	August 2025 (1) (2)
FaceTec	Face	BixeLab	October 2025
Incode	Face	iBeta	February 2026
Oz Forensics	Face	BixeLab	March 2026
Paravision	Face	Ingenium	September 2025
Shufti	Face	iBeta	April 2026
Yoti	Face	iBeta	January 2026

Yet Another Presentation Attack Detection Independent Testing Body

I’ve been compiling a list (April 2 update here) of PAD 3 conforming solutions, or biometric solutions (so far only face) that satisfy a high level of presentation attack detection (liveness detection). This satisfaction is determined by independent testing bodies such as BixeLab, iBeta, Ingenium, and Tüvit.

While nosing around, I found one other entity that performs these assessments: the Swiss Biometrics Center of the Idiap Research Institute.

The SBC’s attestation letters can be found here, including PAD Level A and PAD Level B conformance letters for entities such as Mobai and Identy.

No PAD Level C attestation letters yet.

The April 2, 2026 List of PAD 3 Conforming Solutions

Update to the March 25 version. Added BioID.

Vendor	Modality	Confirming Lab	Link/Date
Aware	Face	BixeLab	November 2025
BioID	Face	Tüvit	August 2025 (1) (2)
FaceTec	Face	BixeLab	October 2025
Incode	Face	iBeta	February 2026
Oz Forensics	Face	BixeLab	March 2026
Paravision	Face	Ingenium	September 2025
Yoti	Face	iBeta	January 2026

I’m slowly finding these vendors. I won’t maintain this list forever, but as long as there are so few Level 3 solutions, I want to highlight them.

Coincidentally, I just reviewed an eBook by one of the vendors listed above, detailing things that you should seek in your liveness detection vendor.

The eBook listed several items.
To no one’s surprise, this particular vendor provided ALL of these items in its liveness detection solution.
Surprisingly, however, the vendor did NOT mention independent confirmation of PAD capabilities.

What is the NIST Facial Recognition Technology Evaluation (FRTE)? And Why Should You Care?

I’m guilty of acronym overuse. I just wrote a post that mentioned something called “FRTE,” and I belatedly realized that many of the people who read the post…and many of the people who need to read the post…have no idea how to spell FRTE, much less WHY it’s important. So let me explain.

But before I explain FRTE, I should explain NIST. It’s the National Institute of Standards and Technology, part of the U.S. Department of Commerce, and it promotes technology standards throughout the country and throughout the world.

Among the many, many, many things that NIST does, it looks at the use of biometrics for identification and classification of individuals, including face. NIST’s face work is split into face recognition and face analysis. While the latter concerns classification of faces (whether the face is real or a presentation attack, the estimated age of the person), the former focuses on individualization.

But I’m not going to talk about FATE today. Let’s focus on FRTE.

Why FRTE?

There are hundreds upon hundreds of algorithms out there that purport to compare a face to another face, or to compare a face to many faces, and indicate the likelihood that the compared faces belong to the same person.

And any algorithm provider can claim that its facial recognition algorithm provides 100.00% accuracy or 99.99% accuracy or whatever.

Or that it can search a trillion record database in 0.1 seconds or whatever.

Perhaps the provider even backs up this claim with published data in which the provider tested its algorithm with 1,000 searches against a 100,000 record database and the algorithm did not make a single error.

Are you impressed?

I’m not.

Anyone can score 100% on a self-test.

But what happens when you are given a test by someone else…closed book…with no answer key?

(And yes, I’m aware of the claims that these independent tests are flawed. So design a better one that more than one algorithm provider supports.)

If you’re looking to buy facial recognition technology, the second best way to evaluate the different facial recognition algorithms is to consult the NIST FRTE tests.

These tests are continuous, with new algorithms usually added monthly.
These tests are complex, measuring umpteen diffferent databases and search types. One or more of these may match your particular use case.
These tests are black box. The algorithm providers send their algorithms to NIST, and they are tested against all the other algorithms on identical setups.

Most importantly, the results of these tests are public, and you can view them yourself. The 1:1 testing is here, and the 1:N testing is here.

Oh, and the tests are listed by the algorithm provider, so if Omnigarde says they’ve been tested by NIST, you can look at the test results and find Omnigarde’s algorithm.

And if Vendor X says its algorithm tested well, but you can’t find Vendor X in the algorithm list, then you need to ask Vendor X which algorithm it’s using.

And if Vendor Y says it’s really accurate, but doesn’t state that the algorithm it uses was NIST tested…ask Vendor Y to prove its accuracy claims.

So that’s FRTE. And if your facial recognition vendor isn’t talking about FRTE…ask why.

On Melanin

If you’re examining a person’s fingerprints, palm prints, face, and irises, you need to understand melanin.

The Cleveland Clinic goes into great detail on melanin, but for now I’m going to concentrate on one item.

There are three types of melanin, two of which affect the skin, eyes, and hair.

Eumelanin. There are two types of eumelanin: black and brown. Eumelanin is responsible for dark colors in skin, eyes and hair. People with brown or black hair have varying amounts of brown and black eumelanin. When there’s no black eumelanin and a small amount of brown eumelanin, it results in blonde hair.

Pheomelanin. This type of melanin pigments your lips, nipples and other pinkish parts of your body. People who have equal parts eumelanin and pheomelanin have red hair.

Melanin obviously affects the coloration of your skin, although some parts of your body (such as your fingertips) may have less melanin than other parts (such as your face).

Concentrating on fingertips and faces (and ignoring irises for the moment), let’s look at a situation where we use an optical mechanism (such as an optical fingerprint reader or a camera), along with available illumination, to photograph fingers and faces of people with varying skin tones.

But what if your entire photographic system is based upon reference materials optimized for light melanin levels? As late as the 1970s, Kodak’s reference materials, called “Shirley cards” after the first model, used to exclusively white people.

In the 1970s, photographer Jim Lyon joined Kodak’s first photo tech division and research laboratories. He says the company recognized there was a problem with the all-white Shirley cards.

“I started incorporating black models pretty heavily in our testing, and it caught on very quickly,” he says. “It wasn’t a big deal, it just seemed like this is the right thing to do. I wasn’t attempting to be politically correct. I was just trying to give us a chance of making a better film, one that reproduced everybody’s skin tone in an appropriate way.”

So hopefully today optical devices are properly capturing fingers, faces, and irises of people at all melanin levels.

Or is this wishful thinking?