February 13, 2017 -
PING AN Tech achieved a high score in the latest tests conducted by the University of Massachusetts’ facial recognition technology benchmark Labeled Faces in the Wild (LFW), scoring a face recognition ratio of 0.9960+/-0.0031.
The company has been developing the face recognition technology over the past three years, with its business units repeatedly using the application.
PING AN Tech’s 13 professional companies and 55 partners all tested the application, including Ping An iLoan 2.0, Ping An Life Insurance, and Shenzhen Social Insurance Wallet.
Based on the in-house usage, PING AN Tech was able to develop a facial recognition solution that delivers rapid identification and a high accuracy rate.
The technology has been applied to 108 use cases involving more than 103 million users, creating a high-precision, easily accessible 100 million-plus face database.
PING AN Tech’s application can be used in a range of high security areas including financial risk control, health insurance, social security benefits programs, railway facility access and airport security.
“Face recognition technology is embedded as a module in many complex business processes,” said Liu Fei, chief scientist of artificial intelligence at PING AN Tech. “It is just one tool among many that should be applied as part of a holistic monitoring system and can only be responsible for its own functionalities within the larger system. When a problem occurs, those not familiar with the technology often attribute the problem to the technology, sometimes confusing the recognition rate with the pass rate or other indicators, leading to misunderstanding. The announcement of the authoritative test results is an effective response to these concerns.”
Established in 2007 by the University of Massachusetts, LFW is used to evaluate the performance of facial recognition algorithms under unconstrained conditions.
As the most widely used evaluation set in the field of facial recognition, LFW attracts a few dozen teams from around the globe including Google, Facebook, Microsoft Research Asia, Baidu, Tencent, SenseTime, Face++ and Chinese University of Hong Kong.
LFW has a total of more than 80 test results, with many of the top ranking methods playing a key role in promoting the development of facial recognition technology.