China develops facial recognition payment system with near-perfect accuracy

With 99.8 percent accuracy, the system will be available to both individual customers and retailers by 2015.
Written by Liu Jiayi, Contributor on

The Chinese Academy of Science (CAS) will launch a face recognition payment app next year, which is slated to land with near-perfect accuracy.

As reported by Science and Technology Daily on Sunday, the Chongqing-based research institute said it has set up the world's biggest Asian face database with more than 50 million records.

The database was created with help from with the University of Illinois, and the National University of Singapore.

According to Zhou Xi, director of the CAS Chongqing Green Technology Research Institute, researchers have developed a unique data collecting technology, which could simultaneously gather facial information from 91 different angles. Researchers were also able enhance the system's learning abilities to function stably and accurately in a dynamic environment.

Zhou said that the system scored the highest accuracy, 99.8 percent, according to a Carnegie Mellon international standard, exceeding its predecessors' best record of 97.6 percent.

The facial recognition system has already been utilized at border controls and attendance machines.

Once the users install the according app on their mobile devices and link it with bank accounts or credit cards, it will take "one second" for the system to identify the payer, Zhou said. 

As for security concerns, Zhou said the special algorithm will be highly reliable, and can easily detect whether the scanned faces are real ones, instead of a held-up photo or video. The system, which is equipped with online learning abilities, will automatically study the changing characteristics of the subject, thus preventing weight changes from affecting the results, for example.

Apart from developing the app for individual customers for online shopping, the institute will also be working with brick-and-mortar retailers and setting up face recognition devices for payment at checkout counters, allowing customers to complete purchases without reaching to their pockets.

Editorial standards