BT/ Apple invents face biometric system for smart glasses wearers to unlock other devices

Paradigm

Follow

Published in

Paradigm

22 min readOct 23, 2023

--

Biometrics biweekly vol. 75, 9th October — 23rd October

TL;DR

Patently Apple reports Apple has filed a new patent application for a new Face ID system designed for smart glasses and, potentially, the Vision Pro
Fime Asia expands Visa accreditation for contactless payment terminal testing
Google’s mDL reaches three more states: Arizona, Colorado, and Georgia.
Biometric presentation attack detection shows maturity in NIST evaluation
FIDO showcases new certifications for remote IDV and bias at Authenticate 2023
FaceTec reveals liveness check milestone, high marks in BixeLab test
Sumsub launches a free tool to detect deepfakes and synthetic fraud
ISO, NIST compliance updates for Proof, Arana, Yoti, authID
Panasonic, academic researchers say data partitioning can reduce face biometrics bias
SMU researchers use AI to build face dataset for AI training
A recent study published in Scientific Reports examines a new method of biometric verification called CompaRR, which records beat-to-beat information embedded in intervals between heartbeats
Ashade introduces a new printable smartcard for digital ID and document verification
Thales and SoyYo expand partnership for passkeys, reported problems for Amazon
Zwipe’s biometric access control cards go global with Schneider partnership
Innovatrics and Nedap partner to create keyless lockers using face biometrics
IDology expands portfolio of gambling services and improves AML, age verification
Single sign-on for US government services getting face biometrics
SendQuick brings Yoti biometrics to Conexa platform
Jumio expands partnership with NextWealth to scale selfie biometrics checks
EU pushes rollout of ETIAS biometric travel system to 2025
New York State called out for fuzziness added to ID proofing regs
Chinese govt limits what biometrics and data can be used to train generative AI
Germany introducing biometric photo standard requirement for ID documents
Kenya’s digital ID boosted by Gates Foundation but govt struggles to convince citizens
Austrians prepare for full switch to new digital ID system Dec. 5
NADRA boss accuses staff of issuing national digital IDs to foreigners
BBC uses facial recognition to identify terrorists as police officer
Japanese AI experts detect Putin's body doubles
Prove secures $40M to expand global digital identity verification
Vietnamese voice biometrics startup Namitech secures $2M funding
Digital identity wallets to reach 1.5B by 2029: Goode Intelligence report
Biometric industry events. And more!

Biometrics Market

The Biometric system market size is projected to grow from USD 36.6 billion in 2020 to USD 68.6 billion by 2025; it is estimated to grow at a CAGR of 13.4% during the forecast period. Increasing use of biometrics in consumer electronic devices for authentication and identification purposes, the growing need for surveillance and security with the heightened threat of terrorist attacks, and the surging adoption of biometric technology in automotive applications are the major factors propelling the growth of the biometric system market.

Biometric Research & Development

Latest Research:

SMU researchers use AI to build face dataset for AI training

Researchers at Southern Methodist University in Texas have been using generative AI to build a demographically rich database of synthetic images of humans to help train facial recognition systems. The images were constructed using the Nvidia DGX SuperPOD platform, and the project being among the first in a recently established “Intelligent Systems and Bias Examination Lab”. Researcher Corey Clark explained that gathering real-world images faces “challenges like consent, fairness, and legal compliance,” whereas synthetic, AI-generated images remove those obstacles.

SMU lab creating large synthetic datasets to reduce bias, preserve privacy in AI facial recognition…

The quality of any artificial intelligence (AI) model relies on the data it is given. That is why researchers at SMU…

www.eurekalert.org

Panasonic, academic researchers say data partitioning can reduce face biometrics bias

Two Panasonic divisions and a Singapore university have developed a way to train face biometrics algorithms that they say improve the performance of facial recognition for demographic groups represented by less training data, EE Times Asia reports.

New Tech Solves Conventional Issues in Facial Recognition Accuracy Surrounding Race and Gender - EE…

Research paper published at International Conference on Computer Vision 2023 lays out new method to enhance accuracy of…

www.eetasia.com

The partner organizations are Panasonic Connect Co. Ltd, Panasonic R&D Center Singapore (Singapore Research Institute) and Singapore’s Nanyang Technological University (NTU).

A paper on the face biometrics training method, which they call “Invariant Feature Regularization for Fair Face Recognition,” has been accepted for publication by the International Conference on Computer Vision (ICCV) 2023.

The researchers say models tend to pick up spurious demographic-specific features. These can be removed through causal intervention, but making the required annotations is prohibitively expensive.

Their method involves generating “diverse data partitions iteratively in an unsupervised fashion,” according to the abstract. The data partitions act as a self-annotation feature to “deconfound” the model through “Invariant Feature Regularization (INV-REG).”

The researchers tested the method against the Masked Face Recognition Challenge evaluation dataset and found that error rates were reduced across four racial groups and images of females.

University of Wisconsin researchers looking into bias in technology and social media suggest that unconscious biases, left unchallenged due to by huge disparities in tech workforces, have contributed to disparate results in a wide range of systems, including facial recognition.

UW researchers untangle bias in technology, social media

Researchers attribute bias, racism in technology to data systems, training sets, algorithms, users

badgerherald.com

Researchers with the U.S. Department of Homeland Security recently found that while race and gender demographic disparities among face biometrics algorithms have improved, they are present among the majority of models.

Heartbeat accuracy for biometric verification shows promise and limitations

A recent study published in Scientific Reports examines a new method of biometric verification called CompaRR, that records beat-to-beat information embedded in intervals between heartbeats.

While many forms of biometric verification involve stationary physiological signals like facial authentication and fingerprint scanning, non-stationary signals like heartbeats are more difficult to fake.

While ECG (or “EKG”) recordings have shown the capacity to verify identity through the heart’s electrical activity, the method uses a complex system that usually requires a medical professional to set up.

By using a method that measures beat-to-beat time intervals, wristbands and video cameras can gather the biometric data needed to verify identity in a more cost-effective, user-friendly way.

The study extracted recordings of heartbeats from 30 mice under anesthesia ranging from 6 to 24 months of age, which is equivalent to roughly 20 to 75 human years. Using the proposed method, researchers were able to verify the identities of the mice within fifty heartbeats, which is close to the resting heart rate for humans with a minimal equal error rate (EER) of 0.21.

For context, the accepted EER rate for most biometric verification methods ranges from 0.1 to 10 percent, the researchers say, depending on the modality and reference database size. ECG-based heart monitoring algorithms have been found to deliver an EER ranging between 0.2 and 19.15 percent. However, the researchers also noted limitations in the reliability of these accuracy evaluations.

The study found that EERs for mice were age-independent until the mice reached the age of 18 months, roughly equivalent to age 50 for humans. Verification was still possible when the model was trained on drug-treated mice, the study says. Verification was less accurate for mice that were awake, possibly due to differences in circadian rhythms throughout different states.

Using beat-to-beat heart signals for age-independent biometric verification - Scientific Reports

Use of non-stationary physiological signals for biometric verification, reduces the ability to forge. Such signals…

www.nature.com

Main News:

Apple invents face biometric system for smart glasses wearers to unlock other devices

Patently Apple reports Apple has filed a new patent application for a new Face ID system designed for smart glasses and, potentially, the Vision Pro.

This innovation would allow Face ID to synchronize with the cameras on and in the head-mounted display (HMD) to unlock companion devices such as Apple Watches, iPhones, iPads and Mac desktop computers.

Apple’s patent addresses the challenge resulting from an HMD, like smart glasses, obscuring a user’s eyes with IR filters. This obstruction prevents the user’s ability to unlock their companion device using facial recognition technology, for instance, if a user attempts to unlock their smartphone using face biometrics while wearing smart glasses equipped with IR filters. Even if they position the phone to capture their face, the IR filters block the facial recognition system from recognizing the user’s eyes, and the device can’t unlock.

The new hardware model overcomes this challenge by using the onboard vision system of the HMD. Based on the location of the companion device and the gaze of the user, whether the user is looking at the companion device and thus intending to unlock the device.

Leveraging the HMD’s vision system, it can determine, based on the user’s gaze and the device’s position, whether the user intends to unlock the companion device. The HMD’s vision system, through either its user-facing or outward-facing camera, gauges the user’s gaze direction and the HMD’s relative position to the device. A gaze oriented towards the device within a narrow range of one to five degrees confirms the user’s intent. The Vision Pro is also going to have iris biometrics capabilities.

Once the system confirms that the user is looking at the companion device, the HMD and the companion device can communicate with each other to exchange authentication credentials for unlocking the companion device.

While the patent emphasizes smart glasses, Patently Apply proposes this could also apply to other HMD devices, such as the Vision Pro.

Fime Asia expands Visa accreditation for contactless payment terminal testing

Fime Asia in Taiwan can now validate Level 2 compliance for the contactless payment terminals of vendors in Visa’s network, after the French company received an accreditation from the payment cards giant.

For Level 2 transactions, businesses collect and process additional data for each transaction from the buyer. Elements such as the customer code, tax amount and tax identification are collected and processed along with Level 1 data such as credit card number and expiration, billing address and zip code.

The accreditation allowing Fime Asia to validate Level 2 compliance for Visa’s contactless terminals is the latest plaudit for the Taipei branch of the France-based firm, which provides comprehensive Level 1, Level 2, Level 3 and biometric testing for payment cards, devices and acceptance terminals across the majority of international and domestic payment schemes.

According to an announcement, the accreditation means that Fime Asia is now positioned to meet all the payments compliance needs within the APAC market. The company says the extension will enable Fime’s customers to ensure testing for their products while spending less time and money.

“We are simplifying and speeding up certification for the APAC market, giving payments industry players the means to create trusted new ways to pay,” says Noël Catherine, vice president of services at Fime.

Fime says the latest accreditation from Visa adds to its two decades of experience offering a wide range of lab and testing services in Asia.

Google’s mDL reaches three more states

Google’s digital driver’s license has come to three more states: Arizona, Colorado, and Georgia. Residents can upload a virtual version of their license or state ID on the Google Wallet app, through a selfie-based process that asks the user to capture photos of the front and back of their ID, and to pose their head in different positions as they take a selfie video. The development comes after Google launched its mobile ID system in Maryland earlier this year.

Google Wallet extends digital ID support to Arizona, Colorado, and Georgia

Foldable phones might miss out on Google Wallet's digital ID feature

www.androidpolice.com

Biometric presentation attack detection shows maturity in NIST evaluation

The effectiveness of the best presentation attack detection (PAD) software for face biometrics is sufficient to give businesses that use it confidence they are catching most spoof attempts, at least based on the latest laboratory testing.

The inaugural evaluation of PAD technologies by the U.S. National Institute of Standards and Technology (NIST) for the Face Analysis Technology Evaluation (FATE) has results several of the leading developers can claim as wins. Perhaps most encouraging for businesses is the improved performance that can be achieved by combining multiple PAD algorithms.

The report shares evaluations of 82 passive, software-based face PAD algorithms operating on conventional 2D images. Each algorithm is submitted for evaluation in preventing impersonation, evasion or both, though in practice, all were implemented for impersonation, and 27 for evasion. The results are broken down into 9 attack types, represented by 16 tables for attacks with still media, and another for stills versus videos. Each is divided into convenience, measured as attack presentation classification error rate (APCER) at bona fide presentation classification error rate (BPCER) 0.01, and security, which is BPCER at APCER 0.01.

ID R&D, Alice Biometrics, Rank One Computing, South Korea-based Kakao Brain, iProov, CyberLink, Onfido, Aware, Saudi Arabia’s STCon and Thai developer Kasikorn Labs (part of Kasikorn Bank Group) each took the top spot in at least one category among the results tables. STCon, CyberLink, Alice, ID R&D, Kakao and ROC.ai each topped both the security and convenience sides of at least one results table.

ROC.ai did not register any false positives or negatives in the evasion test for presentation attack type 6, in one of several encouraging scores.

“Aware applauds NIST for both the rigor and the thought leadership demonstrated in designing this evaluation,” says Dr. Mohamed Lazzouni, CTO of Aware. “Aware strongly believes in building systems in tune with the operational conditions we encounter in our daily lives. As such, providing an optimal experience to balance security and convenience is critical. This is why Aware participated in both impersonation and evasion tasks and performed exceptionally well in both.”

FIDO showcases new certifications for remote IDV and bias at Authenticate 2023

At the 2023 Authenticate conference in Carlsbad, California, a trio of speakers from the FIDO Alliance reflected on the current status of its biometrics requirements and certification program, and offered a preview of new certification for face verification for remote identity verification (rIDV) and identity proofing launching in November.

In a presentation titled “FIDO’s Biometric Journey,” Stephanie Schuckers and Greg Cannon from FIDO’s Biometrics Working Group joined Anthony Lam from its Identity Verification and Binding Working Group for Face Verification to discuss its forthcoming face verification testing for rIDV, which utilizes selfie biometrics matched against the face image on a document.

FIDO launched its document authenticity verification working group last year, and will formally launch the face verification working group in the coming weeks.

Schuckers, vice chair of the Biometrics Working Group, also previewed the FIDO Alliance’s project to develop an optional, add-on certification to address concerns about bias and fairness in biometric technology. Schuckers says that “because this is not a modality thing — fake fingerprint versus face — but an actual implementation, this is really well suited for certification.” The group is currently working to answer questions about the parameters of demographic breakdown, and in Schuckers’ words, “what’s a reasonable requirement for the demographic differentials.”

At present, the system is looking at three age groups, two genders (with ‘other’ as a third option that will not be analyzed due to low sample size) and five different racial or ethnic identity groups.

FaceTec reveals liveness check milestone, high marks in BixeLab test

FaceTec, Inc. now conducts over two billion 3D biometric liveness checks yearly, of which 475 million were performed in Q3 2023, the company has revealed in a review of several new developments. With a year-over-year growth rate of 172 percent in this field, FaceTec emphasized the rising demand for biometric cybersecurity provided by liveness detection.

“Our Q3 results reflect the growing need for effective Liveness-proven biometric cybersecurity. Organizations must defend against generative AI deepfakes to prevent digital impersonation of their employees and end-users using trusted identity reverification with strong, liveness-proven biometrics,” says Kevin Alan Tussy, CEO of FaceTec.

FaceTec’s AI powers biometrics for Utah’s mobile driver licenses (mDLs) through a partnership with GET Group and Scytáles, and is currently working on several other similar implementations.

In recent testing by BixeLab (a NIST-certified biometric testing provider) of FaceTec’s Presentation Attack Detection (PAD), all 2,100 Level 1(A) and 2(B) attacks failed to defeat FaceTec’s defenses.

The PAD testing was performed with 1,000+ live-user, real-world, demographically diverse, 3D liveness checks plus ID scans with 3D:2D face matching. The results showed no observable errors related to race, ethnicity, gender, skin tone, or age, and high 3D:2D match rates at >1/100,000 FAR (false acceptance rate) in most cases, according to the announcement. These test results show that FaceTec’s technology exceeds the levels required for TDIF Accreditation by the Australian Government.

FaceTec also reports a 50 percent year-over-year revenue increase in the third quarter, and more than 170 million facial age estimations performed with its software so far. The company was also granted three more patents, raising its total to 28.

October 2023 marks the fourth anniversary of FaceTec’s Spoof Bounty Program, which continues to be the only biometric bounty program in the world. After over 150,000 attempted attacks, no bounties have been paid out in the last 3.5 years.

Sumsub launches a free tool to detect deepfakes and synthetic fraud

Sumsub has announced the release of For Fake’s Sake, a set of models enabling the detection of deepfakes and synthetic fraud in visual assets. It is the first solution developed by a verification provider and made available to all for free. For Fake’s Sake helps users estimate how likely it is that an uploaded image was created artificially.

Sumsub’s AI/ML Research Lab developed For Fake’s Sake — a set of four distinct machine learning-driven models for deepfake and synthetic fraud detection. This release presents an experimental technical strategy to help people responsibly engage with AI-generated content. This tool also has the potential to adapt and grow with other AI-driven tools. Following this initial contribution, Sumsub will leverage feedback from the AI research community to improve the models’ capabilities further.

“We’ve dedicated substantial efforts to combat the menace of deepfakes and the adverse impacts of synthetic fraud by constantly upgrading our existing solutions and developing new features, including the newly enhanced deepfake detector of our liveness tool,” comments Vyacheslav Zholudev, co-founder and CTO of Sumsub.

According to Sumsub’s most recent internal data, the first half of 2023 saw a considerable outbreak of deepfake cases worldwide compared to the second half of 2022. Deepfakes increased by 84 percent in Great Britain, 250 percent in the U.S., more than 300 percent in Germany and Italy and 500 percent in France. AI-generated identity fraud cases were noticed in Australia (1300 percent), Vietnam (1400 percent), and Japan (2300 percent).

“As AI technologies advance, we foresee a tightening of regulations governing their use. For example, it may soon become mandatory to apply watermarks to all synthetic images. However, fraudsters will continually seek ways to overcome regulations,” explains Pavel Goldman-Kalaydin, head of AI/ML at Sumsub. “We’re pleased to share our experimental synthetic image detection models as a benchmark for further development in the battle against AI-generated fraud.”

ISO, NIST compliance updates for Proof, Arana, Yoti, authID

Arana Security, Yoti, Proof and authID have each announced compliance with data security standards to provide clients with assurance that their handling of biometrics is safe, while a national ID authority has paused third-party enrollments to ensure they meet international standards.

ISO, NIST compliance updates for Proof, Arana, Yoti, authID | Biometric Update

Arana, Yoti, Proof, authID have announced compliance with data security standards to provide clients with assurance…

www.biometricupdate.com

Chinese govt limits what biometrics and data can be used to train generative AI

Data used to train AI should be subject to a security assessment to identify “illegal and harmful information” and organizations should seek consent from individuals whose biometric data is used in training, say new requirements proposed by a committee made up of officials from two Chinese government agencies and the police.

According to a report by Reuters, the draft requirements would see any body of data assessed to have more than 5 percent of said undesirable content blacklisted from being used to train public-facing AI models.

China proposes blacklist of training data for generative AI models

China has published proposed security requirements for firms offering services powered by generative artificial…

www.reuters.com

Among the kinds of information deemed harmful or illegal is anything that advocates terrorism or violence — a standard security concern. But in keeping with Beijing’s tightly-monitored, autocratic security stance, the draft requirements also flag content that involves “overthrowing the socialist system,” “damaging the country’s image,” and “undermining national unity and social stability,” and bans any content that has been censored on the Chinese internet.

Among members of the National Information Security Standardization Committee, which published the draft, are officials from the Cyberspace Administration of China (CAC), lobbying since April to have security assessments for firms developing generative AI intended for public use. Representatives from the Ministry of Industry and Information Technology and the police are also on the committee.

While censorship watchdogs may see another mechanism of government control, the requirements are intended to support China’s drive to rival the U.S. in the AI sector and become a world leader in the field by the end of the decade. As such, they also include detailed guidelines on how to avoid intellectual property violations.

Ashade introduces new printable smartcard for digital ID and document verification

Ashade Tech Inc. has announced the launch of TrustPattern, a technology that enables the storing of information on any surface through image pattern recognition. The technology uses advanced computer algorithms to encode data, creating image patterns that can be easily printed on any surface using standard printers.

TrustPattern supports data compartmentalization and controls access to information, and the company claims it is the only technology that provides dual authentication with fine-grained control of access to information. Additionally, the technology supports cloud, on-site, online, offline and hybrid deployments.

TrustPattern technology is versatile and can be integrated, using their API, into any mobile, web-based, or on-site application. It can be used for digital IDs, financial documents, travel passes, education certificates and other documents that require counterfeit detection.

Dr. Lalitha Kaleedhass, President at Ashade Tech Inc., stated, “The technology can store all the information a smartcard chip stores without requiring smart chips, cards, chip encoders, etc. Hence the name ‘printable smartcard’. This is a first-of-its-kind technology.”