Two days ago, a topic "Billionaire Finds His Son Lost for 25 Years" attracted the attention of the entire Internet. In 1998, Xie Qingshuai was stolen when he was only 3 months old. His father has been searching hard, and finally the father and son family were reunited 25 years later. This time Xie Qingshuai was able to reunite with his family.It is inseparable from a company that does facial recognition.
The company's "cross-age sibling face comparison algorithm" analyzes the facial images of Qingshuai's parents and brother to match Qingshuai, and finally confirms Qingshuai's identity through DNA comparison.
Source: Sina Weibo screenshot
Here comes the question: What exactly is the face recognition technology used to find Jie Qingshuai? Is it the same technology we use to unlock our phones? Today let’s talk about the past and present of face recognition.
From human assistance to pure computer recognition
1. Define a person by physical characteristics
As early as 1879, French criminologist Alphonse Bertillon invented a method for identifying criminals or suspects: identifying a person by measuring 11 physical data, such as the elbow to the end of the middle finger, the length of the right ear, the length of the left foot, etc.
Alphonse's human body feature recognition method, picture source: wikimedia
Although Alphonse's original intention was to identify bad guys, his idea of identifying a person based on body data characteristics has greatly inspired future face recognition technology.
2. “Semi-automatic” face recognition
In 1964, American mathematician and artificial intelligence expert Woodrow Bledsoe and others began to try to use computers to recognize human photos.
The original idea was to take photos of faces from different angles and then let the computer learn the light and dark data on these photos to identify a face. But considering the technical limitations at the time, this idea simply didn’t work.
In the end, Woody and others chose a method similar to Alphonse's, looking for about 20 features from the human face, such as the width and spacing of the eyes, the length of the ears, the length of the corners of the mouth, etc., and defining a person by measuring the values of these features.
However, computers at that time could not directly measure these data from photos, so Woody had to design a software that manually input the measured data into the computer. The computer then compared them with the data in the database to identify who the face belonged to.
This can be regarded as the earliest face recognition software, but it is a "semi-automatic" method that requires human participation and can only process 40 pictures per hour.
3. Fully automatic face recognition
In the 1970s, Japanese scientist Takeo Kanede demonstrated a new facial recognition software. This software can automatically locate the position of the chin, thereby automatically measuring facial data and automatically identifying it.
Although this software will still be affected by factors such as shooting angle, light and shadow, and the recognition accuracy is limited, it has achieved an important transition from "semi-automatic" to "fully automatic".
In the 1980s and 1990s, methods such as Eigenfaces and Fisherfaces appeared, which made significant progress in fully automatic face recognition technology. The recognition ability of static photos (such as ID photos) in controlled environments has become relatively reliable.
For example, West Virginia and New Mexico in the United States have adopted facial recognition technology to identify faces on driver's licenses to prevent the same person from applying for multiple driver's licenses with different names. In 1997, the state of Minnesota in the United States also began using facial recognition systems to identify criminals in the state.
4. Convolutional neural networks bring a leap
Around 2010, deep learning technology based on convolutional neural networks made another leap forward in face recognition technology.
The convolutional neural network is a neural network built to imitate biological vision. Its own architecture is very good at processing image information. People perform various optimizations based on convolutional neural networks and can extract different types of information on the picture as needed, such as light and shade, line contours, etc., to make comprehensive judgments on the picture.
Face recognition system, picture source: Reference [2]
This method is somewhat different from Alphonse's previous method. It no longer identifies faces based on specific numbers such as eye width and ear length, but directly identifies the overall characteristics of the face, which is more similar to the recognition process of animal vision.
Around 2010, computer vision databases that can be used for deep neural network training are becoming increasingly complete. For example, ImageNet built by Li Feifei and others contains tens of millions of manually annotated images. Coupled with the application of GPUs to the field of deep learning around 2012, the speed and accuracy of face recognition technology have been greatly improved.
for example,In 2014, FaceBook's DeepFace facial recognition system has achieved an accuracy of 97.35%, which is almost the same as human recognition accuracy.FaceNet, proposed by Google in 2015, has a face recognition accuracy of 99.63% in some databases.
Deepface face recognition system, picture source: Reference [3]
Today, face recognition technology is still developing rapidly. In addition to identifying still photos, it can also accurately and quickly identify dynamic faces in video images.
Ubiquitous facial recognition technology
As the accuracy of face recognition becomes higher and higher, it is used in more and more applications in life.
For example, almost all mainstream mobile phones now support face unlocking; when buying things in many convenience stores, you don’t need to take out your phone to open the payment code, you can pay directly by swiping your face; when entering or leaving the train station, you can also directly swipe your face to pass through the gate.
In addition to these daily applications, face recognition systems also play an important role in some special fields.
For example, since April 2018, security and surveillance cameras have successfully captured nearly 100 fugitives or criminal suspects at Jacky Cheung's concerts. Netizens also ridiculed Jacky Cheung's concerts as "capture parties."
In addition, in the Lao Rongzhi case, which caused considerable controversy within society,The facial recognition system also easily identified Lao Rongzhi, who had been on the run for 23 years.
Source: Xiamen police video screenshot
As facial recognition technology becomes more and more accurate in cross-age facial recognition and face occlusion, facial recognition technology can also better protect our safety.
The "cross-age and same-kin face comparison algorithm" mentioned earlier is also an optimization of face recognition technology in specific fields. Such a function can not only find absconding criminal suspects many years later, but is also very helpful for missing or abducted children to find their biological parents.
Facial recognition system based on kinship, picture source: Reference [4]
It is worth noting that although AI face recognition brings convenience to our lives and protects our safety, it also has some risks.
For example, there are some apps on mobile phones that use our facial data. Although most companies will try their best to ensure the security of this data,However, there are still some software developers who are unable or have no intention to protect data security, which may cause serious personal information leaks.
For example, ClearviewAI, a company that provides services to U.S. law enforcement agencies, has been exposed to data leaks. In addition, the company was also exposed as illegally collecting user photo information to train algorithms without the user’s consent.
In short, although face recognition technology has a history of more than half a century, it has only really entered a stage of rapid development and penetrated into all areas of life only in the past ten years or so.
Facial recognition technology, used in the right place, can indeed bring convenience and security to our lives, and can also reunite families who have been separated for many years.
But while enjoying the convenience of technological development, we must also enhance our privacy awareness and try to avoid using our own photos and videos on unknown mini programs to reduce the possibility of information leakage.
References
[1]AdjabiI,OuahabiA,BenzaouiA,etal.Past,present,andfutureoffacerecognition:Areview[J].Electronics,2020,9(8):1188.
[2] Wang Liang, Huang Yongzhen, Zhang Kaihao. A method and device for facial kinship recognition based on convolutional neural network, CN105005774A, 2019-02-19
[3]TaigmanY,YangM,RanzatoMA,etal.Deepface:Closingthegaptohuman-levelperformanceinfaceverification[C]//ProceedingsoftheIEEEconferenceoncomputervisionandpatternrecognition.2014:1701-1708.
[4]RobinsonJP,ShaoM,WuY,etal.Familiesinthewild(fiw):Large-scalekinshipimagedatabaseandbenchmarks[C]//Proceedingsofthe24thACMinternationalconferenceonMultimedia.ACM,2016:242-246
Planning and production
This article is a work of Popular Science China-Starry Sky Project
Produced by Science Popularization Department of China Association for Science and Technology
Producer｜China Science and Technology Press Co., Ltd., Beijing Zhongke Galaxy Culture Media Co., Ltd.
Author丨Scientific leftovers science popularization creative team
Review丨Yu Yang, head of Tencent Xuanwu Lab
Planning丨Lin Lin
Editor丨Lin Lin