Hitoshi Imaoka

September 28, 2018
(Updated: May 1, 2019)

Hitoshi Imaoka
Dr. of Engineering
NEC Fellow

Developing the world's fastest and most accurate face recognition *

I work on the research of face recognition technologies at Biometrics Research Laboratories. NEC's face recognition technology ranked No. 1 three consecutive times in still image face recognition tests organized by the U.S. National Institute of Standards and Technology (NIST). It has also ranked No. 1 in video benchmark testing in 2017, demonstrating its high performance.
The key point for face recognition is a robustness that ensures recognition of faces despite any environmental change. In addition to changes in outside light and closed eyes, it is crucial for face recognition technology to tolerate various changes to the face, including different face orientations, without compromising accuracy in order to be useful in real-life applications. In view of that, the question was: how should we process images and what pattern recognition to use? And so we developed the world's best system by building a robust algorithm and leveraging image processing and deep learning. Speed was another critical theme. Today it is possible to compare 30 million pieces of data in about one second. I can safely say that I have spent half my life on the technological evolution on face recognition.
A bit about myself, I studied theoretical physics during my university years. After joining NEC, I engaged in the research of the visual information processing of the human brain, followed by research into face recognition using image processing. Looking back, I feel that all of these three areas of research have been useful to my current efforts. Students and researchers aiming to enter R&D in the area of face recognition may find research in various fields to be valuable input.

  • *
    MBGC 2009 (Multiple Biometric Grand Challenge)
    MBE 2010 (Multiple Biometrics Evaluation)
    FRVT 2013 (Face Recognition Vendor Test)
    FIVE 2017 (Face In Video Evaluation)

Face recognition as the entry point to the next stage

In the early 2000's, face recognition was still unknown to the general public. However, nowadays we see this technology, particularly NEC technology, in use at immigration gates at numerous airports, with which I presume that face recognition has somewhat earned a place in society. I believe that in the near future this technology will find its way into a broader range of uses such as admission control at sports events. As face recognition becomes more common, payment at supermarkets, entry and exit at train ticket gates, and even locking and unlocking offices and homes can be done with your own face. People will no longer need to carry around wallets and keys. We can say that face recognition has tremendous potential.
Also what I would like to draw attention to is the fact that face recognition can be an entry point to communication. When we humans find a friend's face, what do we do? We would probably start a conversation or continue on to different interactions. From such a viewpoint, I would like to pursue technology development beyond that, with face recognition as the entry point.

A society where face recognition replaces existing passwords

What we are aiming for is the realization of a society where face recognition replaces existing passwords. It is a society where people can use their own faces as password to authenticate themselves for entry anywhere anytime. For example, even in a situation where your hands are full with luggage and you cannot take out your keys, you can readily pass doors by just turning your face in that direction. Any and all gates in the world will be passable with only your face as the key. As you can tell by the fact that a passport always has a photo ID, it is a huge advantage that faces can be used universally for identity, with no language barriers. Your face can be used anywhere in the world as the same password that cannot be forgotten.
Currently we are also developing new techniques that use deep learning and the image processing technology that we cultivated for face recognition. In the medical field for example, we developed a technique that supports the discovery and diagnosis of cancer from real-time endoscopic images. Other examples of technological developments utilizing face recognition include users' line-of-sight detection, which can be effectively used in marketing. To make a richer society, we will continue to take on new challenges.

Relevant Laboratories