Yitu Tech ranks first in ID accuracy at Face Recognition Prize Challenge 2017
Chinese facial recognition startup Yitu Tech outperformed more than 12 international firms in face identification accuracy by taking first place in the first Face Recognition Prize Challenge (FRPC), hosted by the Intelligence Advanced Research Projects Activity (IARPA) under the U.S. Office of the Director of National Intelligence.
The victory represents a key milestone for the Chinese AI industry as well as an endorsement of Yitu’s algorithms which recognize faces from images taken under unconstrained and often extreme conditions.
The company’s algorithms overcame challenges posed by complicated environmental conditions, such as non-frontal head poses, poor and uneven illumination, motion blur, and non-neutral facial expression.
The test proceeds by NIST executing participants’ pre-packaged software libraries on images sequestered at NIST.
The test dataset includes “non-cooperative” images of subjects who were either unaware of the camera or didn’t engage with the camera under conditions of overexposure, backlight, profile-view face, and distant photography.
The winner of the identification accuracy challenge is determined by considering measurements of identification accuracy.
Participants are rated based on their False Negative Identification Rate (FNIR) that gives a fixed False Positive Identification Rate (FPIR) no higher than 10^-3.
Algorithms run in a turnkey manner, without tuning, training or other in-the-field specialization, which is significantly different than many academic studies.
“This is a further demonstration of Yitu’s capability and comes on top of our win in Face Recognition Vendor Test (FRVT) this July,” said Dr. Shuang Wu, research scientist at Yitu Tech. “However, having a state-of-the-art algorithm alone is not enough to solve real-world problems, you will need supports from powerful computing infrastructure and sufficiently large dataset. Engineering capability also plays a significant role in performance optimization. Yitu is capable of searching through over a billion records in less than 3 seconds, that showcases our strength in engineering in addition to outstanding algorithms.”
Yitu’s facial recognition technologies have been deployed by China Merchants Bank to enable face-authenticated cash withdrawals at ATMs in several cities already.
The company’s technologies have also being used in wide range of applications in industries spanning public safety, financial services, healthcare, customs and ports, and integrated marketing.