laitimes

From "Aunt Mei" to "Mother of Eight"... Can AI search for people soothe the pain?

From "Aunt Mei" to "Mother of Eight"... Can AI search for people soothe the pain?

Image source: Oriental IC

Looking for most of China, and finally even the scammers are no longer in touch, on December 6, 2021, a hug that was 14 years late moved countless people, Sun Haiyang, who worked hard to find his son for 14 years, experienced countless sleepless nights, and finally stayed up to the end, he is also the prototype of the movie "Dear". After 14 years of suffering, fortunately, it is still a positive result.

However, how many parents are still struggling to find their children, and the day of reunion may be near, or it may be far away.

Just like the recent fermentation of the "Eight-Child Mother of Feng County, Xuzhou", the chains on the neck, the few lost teeth, the vague words... According to the latest official notice, on February 22, the Feng County People's Procuratorate approved the arrest of criminal suspects Sang Mouyu and Shi Mouzhong on suspicion of abducting and trafficking women.

Recently, the Ministry of Public Security decided to launch a special operation to combat the crime of abducting and trafficking in women and children from March 1 to December 31.

Parents who have lost their children, families that have lost loved ones... When having to face this heartbreaking topic, technology presents the power of good, posting street advertisements, a mouth and two legs and a few photos and other "needle in a haystack" method has become obsolete, AI technology, Internet platforms, etc. are becoming the "C position" of the search, and greatly improve the probability of success of the search.

01

24 years of searching for a son, the breakthrough is in face recognition

In the "Mother of Eight Children in Feng County, Xuzhou" incident, in response to the problem that "the woman on the marriage certificate photo of Dong Moumin and Yang Mouxia is not the same person as Yang Mouxia", the public security organs have conducted an investigation and found that Yang Mouxia's recent photos were taken from the Douyin video and circulated to the Internet after retouching, which was different from the actual appearance. At the same time, factors such as skin aging, hair degeneration, liquefaction of fatty tissue and missing teeth are also important reasons for Yang Mouxia's appearance changes. On February 22, the physical evidence appraisal center of the Ministry of Public Security identified the same portrait features as the photo of Yang Mouxia and Dong Moumin's marriage certificate with Yang Mouxia's first marriage photo in Yunnan, the screenshot of Yang Mouxia's online video, the photo of Yang Mouxia's ID card, and the recent photo of Yang Mouxia, which reflected the same portrait characteristics.

"In the past, when we received assistance information about someone who was missing or wanted to find relatives, the usual practice was to compare information, including family descriptions, living environment characteristics, physical characteristics, etc., and then artificial information comparison, find similar people and preliminarily confirm, and then dig out more similar details for comparison, and finally DNA confirmation." Zhang Zhiwei, director of the International Cooperation and Protection Center against Human Trafficking at China University of Political Science and Law and executive director of Beijing ProMed Law Firm, is known as a "lawyer for anti-abduction", and he has helped reunite dozens of families on the road of public welfare anti-trafficking for more than a decade. In his words, the previous search process has no scientific basis, repeatedly experiencing hope and disappointment, "to passively wait for the matching information to appear, the probability of invalid comparison in a large amount of information is very high, the search cycle is also very long, and sometimes some invalid information will become an obstacle on the road to reunion." Zhang Zhiwei confessed to the IT Times reporter.

From "Aunt Mei" to "Mother of Eight"... Can AI search for people soothe the pain?

Source: "Lost"

"In fact, I have never understood in my heart, why is my son lost?" This is a line from the movie "Lost Orphans" starring Andy Lau, based on the experience of a man named Guo Gangtang in Liaocheng, Shandong Province, looking for a son. After 24 years of searching for a son, his 2-year-old lost son had become a teacher when father and son met in July 2021.

In Guo Gangtang's 24-year search for a son, Zhang Zhiwei followed for more than a decade, "The most critical breakthrough is that in June last year, he received a clue to Guo Gangtang's son through the Ali 'reunion' system, although his son has grown up, but through face comparison or judged some biometrics that will not change, and finally confirmed by DNA comparison." Zhang Zhiwei told the "IT Times" reporter that now, face recognition has become a very important method in the process of finding people.

02

Find one out of 1.4 billion people

How to let more people know the detailed missing information? Where are the lost people most likely to be? To whom should I push this lost message? These are the keys to successful searching.

The "reunion" system mentioned by Zhang Zhiwei was first established by Ali programmers, and the biggest role of this system is not to catch human traffickers, but to collect clues about missing children for the first time, and the shorter the missing time, the greater the probability of recovery. As long as the first time to call the police, the reunion system will call sina Weibo, AutoNavi map, Baidu map, QQ and dozens of other apps in a few seconds, to push the message of the child's disappearance to a specific radius centered on the place of the incident. People within a specific range, no matter who they are, can provide information to form a "thousand eyes search".

From "Aunt Mei" to "Mother of Eight"... Can AI search for people soothe the pain?

Image source: AutoNavi map

After more than a decade and twenty years, finding someone from 1.4 billion people may sound like a fantasy, but the "reunion" system has helped recover nearly 5,000 children since it was launched in 2016. "The recovery rate is 98%, and the recovery time is also significantly shortened, with the shortest recovery time being less than 3 hours." In Zhang Zhiwei's view, when effective information is received, the faster and more accurate the matching speed and results, the more it can increase the success rate of the reunion, which is unattainable by the naked eye.

Machines can carry out deep learning and build data models based on more real data, and AI tracing technology based on face recognition will further improve operational efficiency and accuracy, so that more lost families can break the mirror and reunite.

By calling the cameras of any networked device in the world, coupled with face recognition technology, so that the people you want to find nowhere to hide, this is a black technology in the movie "Fast and Furious 7".

From "Aunt Mei" to "Mother of Eight"... Can AI search for people soothe the pain?

In China, there is also such a system, called "Skynet Project", its main use is through face recognition technology and big data technology, the face captured by the camera is recognized, and then compared in the database, the comparison speed is extremely fast, 1 second can "screen" the national population. If a missing person is found and the information is also in the database, it can be located, and the nearby police immediately go to the rescue. If the missing person cannot be found for a while, AI technology can also remember the missing person information and perform cross-age portrait analysis, even after ten years or more, the cross-age portrait model has the ability to find the child.

From "Aunt Mei" to "Mother of Eight"... Can AI search for people soothe the pain?

Through film and television works, Chinese audiences know the famous AMBER Alert in the United States, in fact, there are already more technologically advanced similar police and Internet commercial tracing systems in China.

03

Decades later, "one-eyed identification"

Looking for people, time is measured in years, years, decades, decades, sometimes the first sight is still a child, goodbye is already white hair.

"Cross-age face recognition is a major difficulty in the research of face recognition technology, and our project team has built a cross-time domain face retrieval and analysis system, which has become a key technology for recovering missing persons and trafficked people." The team led by Ma Lizhuang, a distinguished professor at Shanghai Jiao Tong University, won the 2020 Shanghai Science and Technology Progress Special Prize for the project "Character Visual Understanding Technology and Application for Complex Scenes", and he told the IT Times reporter that the face shape of infants and young children changes significantly with age, especially the drastic changes in their faces after more than a decade, which poses a huge challenge to finding relatives. The project team's pioneering cross-age face nucleus technology, which can quickly compare with adult appearances with only a few blurry photos of infants and young children, has so far assisted in retrieving 15 children who have been abducted for more than 10 years.

The "little rat" who once appeared on the CCTV "Waiting for Me" program recognized his parents in the program, and he finally returned to his parents after 18 years of abduction at the age of 3, behind which is the credit of the cross-age face recognition technology of the Malizhuang team.

Cross-age face recognition technology is the result of the cooperation between the team and Tencent Youtu, in the case of only a few blurry photos of the infant stage, the Malizhuang team extracts the characteristics of the person that will not change, learns the natural cross-age change law of the face from the data, and can complete the tens of millions of face retrieval within milliseconds, which greatly improves the accuracy and speed of the matching of the lost population. "Artificial intelligence can solve the problem of precision and speed, and it can also give deterrence to criminals." Mali Chuang said.

In fact, the use of AI to find people is also done by many Internet companies. In November 2018, today's headlines launched the "face to find people" function, users upload photos of the lost, you can directly compare the matching results with the database of the lost people in the background of the headline search, the technology used is through the access to the laboratory's facial recognition technology, the use of AI algorithms, the user uploaded the photos of the lost people and the big data of the lost people in the database to help the lost people return to the family; in 2016, the Baidu AI search platform was launched, and the user uploaded a photo on the platform. As of October 30, 2021, Baidu AI tracing platform has helped reunite nearly 13,000 families, many of which have been separated for more than 20 years or even more than 30 years.

From "Aunt Mei" to "Mother of Eight"... Can AI search for people soothe the pain?

Image source: Baidu

04

AR, multi-modal to help find relatives

"Although DNA identification is the final step in confirming the success of the family search, this is based on the previous face comparison." In Zhang Zhiwei's view, compared with the passive registration of the DNA database, the relevant database information of actively grabbing face information is more comprehensive and more convenient.

Therefore, Liu Jiacheng, member of the National Committee of the Chinese People's Political Consultative Conference and vice chairman of the Beijing Television Artists Association, suggested on the eve of the two sessions of the National People's Congress this year that when kindergartens enter kindergartens and primary schools, all children entering kindergartens and schools should be required to collect DNA information and establish DNA archives, and the files should be handed over to the public security departments for unified storage in the DNA database, with the purpose of effectively cracking down on abduction.

From "Aunt Mei" to "Mother of Eight"... Can AI search for people soothe the pain?

However, it is not that face-to-face comparison can make all the lost people walk home. "Technology such as photo blur, picture repair, and face repair will affect the progress of family searching, and there will still be many pain points and bumps on the road to finding relatives." Zhang Zhiwei said that in addition to the laws and regulations for the abduction of women and children to severely punish, technology can also play a greater role, "for example, the now very hot metacosm can be applied to the field of searching, through the description of the searcher, the use of AR to restore the scene of the previous life of the searcher, can help to find relatives." ”

The process of searching for relatives is a process of cross-verification of a variety of information, including physical appearance, behavior, environment, etc. "We have been studying the whole process of cross-time and space-time multimodal character characteristics understanding technology system, if the photo is relatively blurry, it can provide relatively accurate information through cross-verification and analysis of various factors such as behavior, posture, and scene." Mali Chuang said.

It is undeniable that "technology + search" is releasing more and more potential, but the best outcome should be that the less the application of search technology, the better.

From "Aunt Mei" to "Mother of Eight"... Can AI search for people soothe the pain?

Author/IT Times reporter Pan Shaoying

Edit/Kick Sister

Typography / Ji Jiaying

Photo/ "Lost alone" AutoNavi Map Baidu Oriental IC

Source/IT Times public account vittimes

From "Aunt Mei" to "Mother of Eight"... Can AI search for people soothe the pain?

"IT Times" jointly complained with black cats, collected complaint clues, exposed cases of bad businesses harming consumer rights and interests, and gave full play to the role of media supervision and public opinion guidance. Do not forget every tiny voice, protect every legitimate rights and interests, and build a new environment for harmonious consumption.

Read on