Japanese electronics giant NEC Corp. is working to enhance its already state-of-the-art facial recognition technology and further develop its crowd behavior analysis technology to provide safe and efficient security services at the 2020 Tokyo Summer Olympics and Paralympics.
A new way of managing the entrance of large numbers of people was tested out at a sports festival, open to the public, held in mid-November in the Yumenoshima stadium, in Tokyo's Koto Ward. At the entrance of a relay marathon venue, participants were instructed to face the camera of a tablet device that had been set up. Within seconds, the participants' identities were confirmed through software that verified that their facial data, which had previously been collected, matched the facial features taken by the tablet camera.
Some 300 event participants across two days were eligible for the test run of this technology, and all were able to get through without trouble.
The facial recognition system that was used for this test is NeoFace. NEC has one of the world's top verification systems, which has been used for security at the national level including for passport control, and to manage entry into amusement parks. The event in Yumenoshima was the first time the technology was used at a sporting event. As a Tokyo 2020 Gold Partner sponsor of the Olympic and Paralympic Games, NEC is hoping to see its verification and crowd-control technology deployed at the Games.
Shunichi Ueda, manager of NEC's Corporate Communications Department, pointed out that on the two days that the company's facial recognition system was tested, the weather had been cloudy. "Sunlight may affect the system's ability to verify facial data. And because the number of people who will be using the system is expected to become much greater, we need to test it under a wide range of conditions."
In addition, Ueda said, "Our current facial recognition system is the kind where one must stop and face the camera, but we are hoping to develop a 'walk-through' system that will recognize moving facial profiles."
The entry-management system used at the sports event is much faster and more efficient than if event staff were to check participants' IDs visually. The system can also prevent an unwelcome person using someone else's pass to get into an event much more efficiently than visual checks by humans can. The system is already in use at entry checkpoints at pop concert venues. By cross-checking the data of fan club members against people trying to get into concerts, the system can also help prevent ticket scalping. The system has reduced the time needed for identity verification by a maximum of 30 percent, compared to manual ID photo checks.
Is there any risk of personal information leaking, or of people's privacy otherwise being violated by such a system? "Because the system converts facial characteristics into data, the images themselves are not saved on the tablets," Ueda explains. "We've implemented measures to prevent images from being recreated using that data."
Meanwhile, NEC is in the midst of developing crowd analysis technology that would make it possible to detect abnormal behavior. The technology was tested alongside NEC's facial recognition technology at the recent event.
Two cameras were set up -- one at the check-in counter for a reception party and another in the space where participants dined -- and a monitor showed what was taking place. Whenever the number of people in one location exceeded a certain density, the monitor showed that crowded space as red, allowing system users to identify the congestion at a single glance. The aim is to analyze which areas become crowded next, and make it possible to predict such phenomena.
Tokyo's Toshima Ward has been using NEC's crowd analysis technology since September this year. In the wake of the March 2011 Great East Japan Earthquake, an unexpectedly large number of people unable to get home flooded Ikebukuro Station and the surrounding area. The ward office could not get a comprehensive picture of the situation. Learning from this experience, the ward decided to incorporate crowd analysis into reinforced disasters response measures, setting up 51 cameras near major train stations and roads to collect real-time information in times of emergency. The footage collected by these cameras will be analyzed to predict crowding, and an alarm is set to go off if more than a certain number of people gather in a single space.
NEC says its goal is to use computer analysis to develop a system that can predict crowding. Said Ueda, "If we set up robots equipped with crowd analysis technology near sports venues and other places that are likely to become congested, we can direct spectators and passersby to less crowded train stations or areas, contributing to the dispersal of crowds and overall security."