You Can Measure High quality, However is the Measure Significant? (OFIQ) – Bredemarket

The aim of measuring high quality shouldn’t be for measurement’s personal sake. The aim ought to be to tell individuals to make helpful selections.

In Germany, the Bundesamt für Sicherheit in der Informationstechnik (Federal Workplace for Data Safety) has developed the Open Supply Face Picture High quality (OFIQ) commonplace.

Skilled biometric professionals can’t assist however discover that the acronym OFIQ is just like the acronym NFIQ (utilized in NFIQ 2), however the latter refers back to the NIST FINGERPRINT picture high quality commonplace. NFIQ can also be open supply, with contributions from NIST and the German BSI, amongst others.

However NFIQ and OFIQ, whereas analyzing totally different biometric modalities, serve an analogous goal: to tell apart between good and dangerous biometric photographs.

However do these open supply algorithms meaningfully measure high quality?

The examine of OFIQ

Biometric Replace alerted readers to the November 2025 examine “On the Utility of the Open Supply Facial Picture High quality Instrument for Facial Biometric Recognition in DHS Operations” (PDF).

Notice the phrases “in DHS Operations,” that are essential.

• The DHS doesn’t care about how ALL facial recognition algorithms carry out.
• The DHS solely cares in regards to the facial recognition algorithms that will doubtlessly use.
• DHS doesn’t care about algorithms it could by no means use, corresponding to Chinese language or Russian algorithms.
• In actual fact, from the DHS perspective, it most likely hopes that the Chinese language Cloudwalk algorithm performs very badly. (In NIST assessments, it doesn’t.)

So which algorithms did DHS consider? We don’t know exactly.

“A complete of 16 business face recognition methods had been used on this analysis. They’re labeled in diagrams as COTS1 via COTS16….Every algorithm on this examine was voluntarily submitted to the MdTF as
a part of on-going biometric efficiency evaluations by its business entity.”

Often MdTF rally individuals aren’t disclosed, until a participant discloses itself, like Paravision did after the 2022 Biometric Know-how Rally.

“Paravision’s matching system alias within the take a look at was ‘Miami.’”

So what did DHS discover when it used OFIQ to guage photographs submitted to those 16 algorithms?

“We discovered that the OFIQ unified high quality rating gives extraordinarily restricted utility within the DHS use circumstances we investigated. At operationally related biometric thresholds, biometric matching efficiency was excessive and probe samples that had been assessed as having very low high quality by OFIQ nonetheless efficiently matched to references utilizing a wide range of face recognition algorithms.”

Or in human phrases:

• Photos that yielded a top quality OFIQ rating precisely matched faces utilizing the examined algorithms.

• Photos that yielded a low high quality OFIQ rating…STILL precisely matched faces utilizing the examined algorithms.

So, not less than in DHS’ case, it is not sensible to make use of the OFIQ algorithm.

Your mileage could fluctuate.

If in case you have questions, seek the advice of a biometric product advertising skilled.

Or Will Smith. Simply don’t make a joke about his spouse.