How China's Pay By Facial Recognition Works

By Charles Rollet, Published on Apr 02, 2019

Many social media posts have variously celebrated or warned about the growing use of facial recognition for payments in China. An example of one such video is shown to the right.

When one looks at such videos, an obvious question is: how do they do that accurately? There are many millions of people in China and, in these videos, the person seems to simply look at the camera and is promptly recognized.

However, the system is not omniscient. While facial recognition for payments in China is real, it relies on verificationnot identificationa key difference.

In this post, we examine how facial recognition payments really work in China, and how it fundamentally relies on the country's system of national ID cards.

How It Works

In many convenience stores and restaurants in China, it is indeed possible to pay without bringing a cellphone or cash. But it is not as simple as just looking into a camera at the check-out counter. There are actually two steps:

  1. Look into the camera
  2. Enter your cellphone number on the screen

The cellphone number is used to bring up your personal details and an existing photo of your face. That existing photo is compared to the one just taken, and if it is a match, the payment goes through.

Why It Is Possible: Alipay

Alipay is the mobile payments arm of Chinese e-commerce giant Alibaba [link no longer available](in China similar to Amazon in the US). Over 900 million, mostly Chinese people, have Alipay accounts, which require a national ID card, bank account, and mobile phone to activate. Alipay users can also enable facial recognition by taking a selfie with their cellphone. Therefore, just by plugging in a phone number at check-out, Alipay can directly access a high-res image of a user's face along with bank account details. 

Alipay released a video showcasing how this system works here

Get Notified of Video Surveillance Breaking News
Get Notified of Video Surveillance Breaking News

As the video shows, a phone number is required, not just a face. The still below demonstrates:

The video also claims that the camera is able to correctly match users despite heavy make-up, different hair color, and other people being present in the photo, as the still below demonstrates:

This is certainly possible, though the more a facial recognition system allows variance in appearance, the greater the risk that it could mismatch to another person. However, since they are doing verification, not identification, the risk is reduced.

Verification vs Identification

Therefore, payment via facial recognition in China is not done by comparing someone's photo to mass numbers of people in an existing database. Technically, this is called '1:N' or one-to-many identification. This is the kind of facial recognition that would need to be done by video surveillance cameras in shopping malls looking for shoplifters or in cities looking for terrorists, etc.

One example of this somewhat science fiction approach is Dahua's 2017 marketing video where they (impossibly) generate a facial recognition match of a person with a hat, in a shadow, looking away from the camera:

While looking directly into a camera's FoV improves image quality vs public surveillance cameras, identification is still difficult to do accurately and reliably when comparing to even thousands of people and would be vastly more difficult in trying to match across millions or hundreds of millions in China. People look alike, especially across large groups and especially when the system is trying to tolerate differences in hairstyle, makeup, weight gain, etc. 

Rather, in the China payments case, '1:1' verification is being used: Alipay's software is only comparing the photo taken at check-out to an existing photo that is part of someone's Alipay account. As such, the system just needs to assess how probable the face presented at the camera is to the reference image on file.

National ID Card System Crucial

A major factor making such a system easier is that Chinese citizens, by law, must have a detailed national ID card, officially called a Resident Identity Card or  居民身份证. These detailed cards are issued by police and function as a sort of domestic passport. Because a national ID card is required to activate Alipay, an Alipay account is effectively as trustworthy as a legal ID. 

(In the US, however, many people have a driver's license as their only official ID. Because they are relatively easy to get and commonly faked, driver's licenses from many states are not enough to even fly domestic. )

Privacy Concerns

While a national ID card system that is used and shared with corporations for payments adds convenience, it does raise privacy concerns about the data being shared. For example, the ACLU has spoken out against national ID cards expressing concerns about them being used to expand surveillance. Of course, inside China, this is the norm.

Other countries do have national ID card systems, like France. However, in France, having such a card is not obligatory and many people do not have one. Compulsory national ID cards have a controversial past in France, as they were instrumental in deporting Jews to camps during WW2; when France attempted to make biometric ID cards compulsory in 2011, the proposal was heavily criticized and ultimately dropped.

Partly because of the lack of such a history, China is well-known for giving short shrift to privacy issues. For example, the CEO of Baidu (China's Google) once said that Chinese people care less about privacy. Still, there are signs this is changing. China passed a personal privacy law in 2018 described by some as "more onerous" than the EU's GDPR. Alipay, for its part, claimed in its promo video that:

We don't store data. We also protect data by using encryption. We want to do everything in our capacity to protect users' privacy and security online.

Popularity in China

IPVM understands that while facial recognition payment is possible in many Chinese outlets, most people still prefer to use their cellphone. The way this is done is by scanning the unique Alipay barcode that is part of their Alipay 'wallet' at check-out. This does not require inputting a cellphone number or posing for a picture and is therefore significantly faster.

However, if one was to leave a cellphone at home, Alipay's facial recognition feature does enable payments without needing anything at hand, a significant convenience.

UPDATE: Face Pay for ATMs

A LinkedIn post shows a man using facial recognition instead of a card at a China ATM. "No card needed anymore, just bring your face", the poster claims.

Screen Shot 2019-04-04 at 4 01 09 PM

IPVM has confirmed that some ATMs in China do use facial recognition. However, this is the same system as using your face to pay for groceries, i.e. it is based on verification not identification. What the LinkedIn post does not show is that a phone number or national ID number is typically required as well.

5 reports cite this report:

China Subway Facial Recognition System Examined on Jun 24, 2019
A China city of 6+ million people has installed facial recognition-enabled...
China Jaywalking Facial Recognition Guide on May 27, 2019
News reports touting the PRC's AI prowess often showcase facial recognition...
Covert Facial Recognition Using Axis and Amazon By NYTimes on May 20, 2019
What if you took a 33MP Axis camera covering one of the busiest parks in the...
19 Facial Recognition Providers Profiled on Apr 23, 2019
IPVM interviewed 19 facial recognition providers at ISC West to understand...
UK Camera Commissioner Calls for Regulating Facial Recognition on Apr 15, 2019
IPVM interviewed Tony Porter, the UK’s surveillance camera commissioner after...
Comments (5) : Members only. Login. or Join.

Related Reports

Exacq "Facial Matching" Facial Recognition Tested on Sep 03, 2020
Exacq is claiming "Accurate Recall of Every Noteworthy Person of Interest"...
The US Fight Over Facial Recognition Explained on Jul 08, 2020
The controversy around facial recognition has grown significantly in 2020,...
Herta Facial Recognition Plus Masks Tested on Aug 19, 2020
Masks increase face recognition errors, but facial recognition developer...
Face Masks Increase Face Recognition Errors Says NIST on Aug 04, 2020
COVID-19 has led to widespread facemask use, which as IPVM testing has shown...
Verkada Video Analytics Tested 2020 on Oct 08, 2020
In 2019, Verkada released people and face analytics but our testing then...
Facial Recognition: Weak Sales, Anti Regulation, No Favorite, Says Security Integrators on Jul 07, 2020
While facial recognition has gained greater prominence, a new IPVM study of...
Bias In Facial Recognition Varies By Country, NIST Report Shows on Jul 15, 2020
While many argue that face recognition is inherently racist, results from one...
Injes Tiny Temperature Terminal Tested on Jul 17, 2020
While temperature terminals have trended bigger, the Injes DFace801 is...
Beware Rigged China Fever Cameras on Sep 08, 2020
Many China fever camera manufacturers have rigged algorithms dynamically...
Briefcam Responsible Use Examined on Aug 24, 2020
While mega-companies Amazon, IBM, and Microsoft have been criticized for...
Mobile Access Control Usage Statistics 2020 on Sep 21, 2020
Most smartphones can be used as access control credentials, but how...
Face Detection Shootout - Dahua, Hanwha, Hikvision, Uniview, Vivotek on Jul 30, 2020
Face detection analytics are available from a number of manufactures...
China General Chamber of Commerce Calls Out US Politics on Sep 24, 2020
While US-China relations are at an all-time low, optimism about relations...
Uniview H1 2020 Financials Examined on Sep 08, 2020
While Dahua and Hikvision, helped by fever camera sales, are recovering from...
Gait Recognition Examined on Sep 14, 2020
Facial recognition faces increasing ethical and political criticisms while...

Recent Reports

Recruiters Online Show LIVE Today! on Oct 29, 2020
IPVM's 7th online show resumes today with 12 recruiters presenting themselves...
Hikvision AcuSense G2 Camera Test on Oct 29, 2020
Hikvision has released their next generation of AcuSense analytic cameras...
Biggest Problems Selling Access Control 2020 on Oct 29, 2020
Access control can cause integrators big headaches. What practical issues do...
Taiwan Geovision AI Analytics and NDAA Examined on Oct 29, 2020
Taiwan manufacturer Geovision's revenue has been falling for years. However,...
Bedside Cough and Sneeze Detector (Sound Intelligence and CLB) on Oct 28, 2020
Coronavirus has increased interest in detecting symptoms such as fever and...
Fever Tablet Thermal Sensors Examined (Melexis) on Oct 28, 2020
Fever tablet suppliers heavily rely on the accuracy and specs of...
Verkada Fires 3 on Oct 28, 2020
Verkada has fired three employees over an incident where female colleagues...
Eagle Eye Networks Raises $40 Million on Oct 27, 2020
Eagle Eye has raised $40 million aiming to "reinvent video...
Hikvision Q3 2020 Global Revenue Rises, US Revenue Falls on Oct 27, 2020
While Hikvision's global revenue rises driven by domestic recovery, its US...
VICE Investigates Verkada's Harassing "RawVerkadawgz" on Oct 26, 2020
This month, IPVM investigated Verkada's sexism, discrimination, and cultural...
Six Flags' FDA Violating Outdoor Dahua Fever Cameras on Oct 26, 2020
As Six Flags scrambled to reopen parks amid plummeting revenues caused by the...
ISC Brasil Digital Experience 2020 Report on Oct 23, 2020
ISC Brasil 2020 rebranded itself to ISC Digital Experience and, like its...
Top Video Surveillance Service Call Problems 2020 on Oct 23, 2020
3 primary and 4 secondary issues stood out as causing the most problems when...
GDPR Impact On Temperature / Fever Screening Explained on Oct 22, 2020
What impact does GDPR have on temperature screening? Do you risk a GDPR fine...
Security And Safety Things (S&ST) Tested on Oct 22, 2020
S&ST, a Bosch spinout, is spending tens of millions of dollars aiming to...