Dialogue with the champion team of the "Face Recognition World Cup"

Chinese Academy of Sciences "Smart Aviation Safety" exhibition area. Respondents provided pictures Hualong.com

Hualong.com at 10:04 on August 25th (Chief reporter She Zhenfang and intern Tang Lilin) won the crown at the "Face Recognition World Cup" held by Microsoft in 2016. The product was promoted to 70 airports across the country, and it took less than three years for the Intelligent Security Technology Research Center of the Chongqing Research Institute of the Chinese Academy of Sciences.

The reporter learned that based on its self-developed face recognition technology and other intelligent technologies, the center has jointly developed and promoted a number of products in the civil aviation field with Wisdom Air Safety (Beijing) Technology Co., Ltd.

For example, the smart brain of the civil aviation airport uses artificial intelligence to comprehensively analyze the big data of the airport, helping the airport to deploy manpower and material resources more scientifically; promote the security check face recognition auxiliary verification system applied to 618 security check channels in 70 airports, and the original The inspection time of more than 15 seconds is simplified to 2 seconds; the automatic image judgment system of the X-ray security inspection machine solves the "pain point" that each security inspection site needs to spend 1-2 years training a security inspector...

Flowers and applause are always accompanied by challenges. Speaking of the stories behind the team's achievements, Shi Yu, director of the Intelligent Security Technology Research Center of the Chongqing Research Institute of the Chinese Academy of Sciences, told reporters that challenges have always been there.

"In October 2016, the Civil Aviation Administration cooperated with us in order to optimize the security inspection mode and innovate the security inspection process," Shi Yu recalled. At the beginning of the cooperation, the center conducted a two-month test in four airports including Chongqing and Shanghai.

On December 10, 2016, the second meeting of the China Civil Aviation Expert Group was held. More than 20 experts from the Civil Aviation Administration and related airports discussed in detail the applicability of face recognition in civil aviation airports based on the previous test data. It is decided to immediately start the civil aviation security inspection face recognition and witness comparison project. The system equipment is developed and provided by the Chongqing Research Institute of the Chinese Academy of Sciences. After three months of hard work by the central team, the prototype of the first-generation product "Security Inspection Face Recognition Auxiliary Verification System" was completed.

But the top-level algorithm will also encounter many environmental problems in practical applications.

With the massive flow of people at the airport, the requirements for the face recognition system are accurate, more accurate, faster and faster. In order to achieve a better product effect, the central team stayed at the airport, pondered over and over again, and carried out multi-party communication with experts and scholars. The product was upgraded iteratively, even once a month, and replaced every problem.

Finally, on May 1, 2017, the "Security Check Face Recognition Auxiliary Verification System" was successfully unveiled at Chongqing Jiangbei Airport. Since the product was launched, the effect has been very good. In the past, it took about 10 seconds for airport security staff to check a passenger, but after adopting the system, it only takes one second to read the ID card information, "swipe face" comparison and get the result. In addition, the accuracy rate of the staff's naked eye comparison is about 80%, and the security check face recognition auxiliary verification system can exceed 98%, which greatly improves the accuracy rate.

"Especially during the Spring Festival this year (2018), Jiangbei Airport transported a total of 4.87 million passengers, with an average of 122,000 passengers per day and a maximum of 135,000 passengers per day. After using the system, the pressure on the staff has been greatly relieved." Jiangbei Airport staff According to the introduction, in the nearly one year since the security check face recognition auxiliary verification system was launched, they have seized a total of 233 incidents of attempting to board the plane with other people's certificates, which has further improved the security level of the airport.

The team's efforts have brought about results - in the airport scenario, the security check face recognition auxiliary verification system has become the absolute first in terms of face capture, accuracy rate, and pass rate.

In May 2017, Guangzhou Baiyun Airport conducted a large-scale "PK" of intelligent identification technology, and conducted field tests on 4 manufacturers. "During the 50-day test, the test effect of our product is the best." Shi Yu introduced.

Talking about the direction and goals of the team's future efforts, Shi Yu said, "We will deepen cooperation with civil aviation, promote the transfer and transformation of Chinese Academy of Sciences technology in all aspects of the civil aviation field, and contribute more to promoting the construction of a strong country in science and technology and civil aviation."