User:Rambler930/sandbox

The facial recognition system in China is a system that involves facial recognition technology (FRT) which employs in the People's Republic of China. "In China, facial recognition technology — biometric computer applications that automatically identify an individual from a database of digital images — is a part of daily life". The facial recognition technology is a part of the biometric technology and is also a part of the surveillance system. Koc and Barkana described that “a facial recognition system is for automatically identifying or verifying a person from his/her digital image. It is typically used in security systems". "In China, these systems are used in safe cities projects in production". According to N. U. Bagrov et al, the face recognition system includes "the following components: Face detection module, data storage and load balancer, and face recognition module". "Facial recognition technology provides a sophisticated surveillance technique that can be more accurate than the human eye".

History

The development of face recognition technology (FRT) in China started at the end of the 1990s, and 2014 is the key year for in-depth learning to be applied to face recognition. This year, the founder of Megvii, and the Tang Xiao'ou’s team of the Chinese University of Hong Kong achieved excellent results in the field of in-depth learning combined with face recognition. Since 2015, China has put forward a series of development plans and policies to promote the development of information technology and artificial intelligence in China, and face recognition technology ushered in a period of rapid development.

Facial recognition technology is "big business in China, with the country bidding fair by the end of 2018 to constitute 46 percent of the US$17.3 billion global video surveillance market. Surveillance sales to Chinese police by SenseTime, currently China’s largest FRT firm, account for a third of its overall business. Megvii, China’s second-largest FRT firm, achieves rapid technological development because of enormous governmental demand for constantly improving FRT". "Megvii is known for its open-source facial recognition platform, called Face++, which more than 300,000 developers are currently using to build their own face-detection programs". "In China, uses for FRT are mainly, but not entirely, for state surveillance. It is also increasingly used in cab-hailing, security lock systems for doors, and even in commercial venues, where it supposedly gleans data from faces to extrapolate on what items consumers may be apt to purchase". "A Chinese FRT system named Sky Net can theoretically scan the entire Chinese population of 1.3 billion people within one second, even under poor visibility conditions and with the people under surveillance in motion, as in cars or escalators. Its developer claims it has an accuracy rate of up to 99.8 percent".

Technologies

Face recognition technology is a biometric recognition technology, it is based on the facial feature data. "A facial recognition system uses biometrics to map facial features from a photograph or video. It compares the information with a database of known faces to find a match".

Facial recognition technology "uses a software application to create a template by analyzing images of human faces in order to identify or verify a person's identity". "It creates a 'template' of the target's facial image and compares the template to photographs of preexisting images of a face(s) (known)". "The face is measured through specific characteristics, such as the distance between the eyes, the width of the nose, and the length of the jaw line".

Sources of Data

Face Dataset/database

"The known photographs are found in a variety of places, including driver's license databases, government identification records, mugshots, or social media accounts, such as Facebook". In China particularly, the data comes from the resident identity card, Sina Weibo, WeChat, QQ, etc, and also comes from video shooting, camera monitoring, social network images. There are also face databases, the images are collected from the internet or captured by the database laboratories. Examples are:

Cas-Peal

It was collected and established by the Chinese Academy of Sciences. It is mainly to provide a large-scale Chinese face dataset for training and evaluating the algorithm corresponding to Oriental people. At present, CAS-PEAL face database is composed of "99,594 images of 1,040 individuals (595 men and 445 women)". It has different postures, expressions, lighting conditions, expressions, with or without glasses and other information in a specific environment. For each person, nine cameras are used to capture images of different poses at the same time. On average, about 900 images are collected by each person.

FaceWareHouse

Released in 2014, FaceWareHouse is an open-source 3D face dataset from Zhou Kun Laboratory of Zhejiang University, similar to the construction of 3DMM dataset, the dataset is about Chinese people. A total of 150 people, ages from 7 to 80, were included. Compared with the 3DMM dataset, it adds expressions, each person contains 20 different expressions with one neutral expression, 19 mouth opening, smile and other expressions.

CASIA WebFace

It is developed by Stan Z. Li laboratory from the Chinese Academy of Sciences (CASIA) in 2014, it includes 494,414 images of 10,575 people.

Celeba

It is published in 2015, it is a large face recognition data set released by Professor Tang Xiao'ou's Laboratory of the Chinese University of Hong Kong. The "dataset contains 202,599 face images of 10,177 celebrities". There are more than 40 kinds of face attributes, including whether with or without glasses, whether with smiles or not, etc. It is mainly used for "face attribute recognition".

Dragonfly Eye Database

It is developed by Yitu Technology. It is a platform that has 1.8 billion photographs to work with. These photographs come from all the people in the national databases and from all those people who enter and exit China. 320 million of the photographs have come from ports, airports and borders.

Applications

In China, "where mobile payment is already one of the most advanced in the world, customers can make a purchase simply by posing in front of point-of-sale (POS) machines equipped with cameras, after linking an image of their face to a digital payment system or bank account".

Online:

Alipay (by Alibaba)

Alipay is "the financial arm of e-commerce giant Alibaba". Alipay's face payment is co-developed by Ma Yun and Megvii. It can cope with complex illumination and support multiple face gestures. "Alibaba gives customers the option to pay using their face at its Hema supermarket chain". It is a new payment method; by using biometrics technology, it enables the users to pay without turning on their mobile phones. This technology adapted in facial payment effectively improves efficiency and the buyers’ consumption experiences.

ZAO

Launched on "August 30, 2019, ZAO became the most downloaded app in China's iOS App Store over the weekend". It is a free deepfake face-swapping phone application developed by the mobile app giant Momo in 2019. It enables you "to place your likeness into scenes from hundreds of movies and TV shows after uploading just a single photograph". This process only takes a few seconds.

Uses of Mobile Phones

"China's telecom operators must now use facial recognition technology and other means to verify the identity of people opening new mobile phone accounts". "The measure, described by the ministry of industry and information as a way to 'protect the legitimate rights and interest of citizens in cyberspace', makes Chinese mobile phone and internet users easier to track".

Offline:

Subway

Facial recognition faregate at Beijing West Railway Station

Exit turnstile with Face Scanner in Shenzhen Railway Station

"The metro systems of some major Chinese cities have announced they will use" FRT, "with government-owned newspaper China Daily saying Beijing will use it to 'classify passengers' to allow for 'different security check measures'”. Also, Shenzhen, China’s tech capital, is testing the same in a local subway. The passengers enter the station by scanning "their faces on the screen where they would normally have tapped their phones or subway cards. Their fare then gets automatically deducted from their linked accounts. They will need to have registered their facial data beforehand and linked a payment method to their subway account".

Hotel

"Alibaba runs a hotel in its headquarters city of Hangzhou where guests can scan their faces with their smartphones for advance check-in".

Shopping street in Wenzhou city

It is "a historical shopping street in Wenzhou city" and is also "China’s first facial recognition payment-based shopping street. The city's iconic Wuma (Five-horse) Street took priority in expanding the innovative way of 'paying with your face'. Some 20 stores along the street have been equipped with the Alipay system called Dragonfly. At a local store, customers can look at the iPad-sized Alipay device to complete payment in less than 10 seconds".

Public washrooms

FRT is being used in the public washrooms as well. In order to prevent toilet paper theft from taking public toilet paper home (because toilet paper is not readily available in most public washrooms in China), a public washroom at Temple of Heaven in Beijing has adopted "a camera-equipped dispenser for users to receive a two-foot section of bathroom tissue. Users must stand for three seconds in front of the dispenser to get the tissue, and it will not dispense any more to the same user for the next nine minutes".

Banks

Banks in China now are adapting facial recognition technology to check users' identities. "At the Agricultural Bank of China's (ABC) three outlets in Jinan City, cards are no longer needed to withdraw money. A quick scan of the face will do. All you have to do is to press the facial recognition withdrawal button, scan your face in the camera, enter your phone number or ID number, and enter your transaction amount and password". Several other banks also have similar technologies.

Security of housing

FRT has been used for the security of housings in China. In July 2019, "the Xinhua news agency said Beijing had, or was in the process of, installing facial recognition systems at the entrances of 59 public rental housing communities".

Surveillance:

Police

"FRT has the potential to be a useful tool in crime-fighting by identifying criminals who are captured on surveillance footage, locating wanted fugitives in a crowd, or spotting terrorists as they enter the country". “On April 7, 2018, Chinese police in Nanchang, Jiangxi province, successfully used facial recognition technology to pinpoint and nab a criminal suspect out of a crowd of 60,000 people attending a Cantopop concert. FRT had identified the man as he filed through a stadium entrance”. Chinese polices also use facial recognition glasses to catch criminals. "The eyeglass-mounted camera is equipped with facial-recognition technology capable of 'highly effective screening' of crowds for fugitives traveling under false pretenses". "The glasses, which link to a database that matches travelers with criminal suspects, helped identify seven suspected criminals and 26 others using fake identity documents".

Yitu Dragonfly Eye Intelligent Security System

“Yitu Dragonfly Eye Intelligent Security System is built upon machine vision technology to operate an intelligent security system designed for safe cities, intelligent transportation, and smart finance”. It is a generic portrait platform using facial recognition technology, it "can identify a person from a database of at least 2 billion people in a matter of seconds". It was developed by the CEO of Yitu TechnologyLeo Zhu and Lin Chenxi in 2017. "Dragonfly Eye is presently used by 150 municipal public security systems and 20 provincial public security departments across the country of China. Local police authorities credit Dragonfly Eye with aiding in the arrest of 576 suspects on the Shanghai Metro in the first three months of using the facial recognition system".

KFC Stores

Customers are able pay with the facial recognition system in KFC stores in China. “Combined with a 3D camera and liveness detection algorithm, Smile to Pay can effectively block spoofing attempts using other people’s photos or video recordings and ensure account safety”. They work with Ant Financial under Alibaba for this facial recognition system. “Diners can pay by scanning their faces at an ordering kiosk and entering a phone number - which is meant to guard against people cheating the system”.

Forbidden city tourist attraction

In 2019, Chinese telecoms giant Huawei "finalized a deal to build a smart 5G network mobile comms network covering the 720,000 square-metre Imperial Palace (the Forbidden City) compound in Beijing and the accompanying Palace Museum. Huawei will provide the museum with facial recognition sensors and other AI systems with 3,000-plus video cameras installed throughout the building. The FRT will be used to secure the 1.86 million relics housed at the museum, which logged 17.5 million visits from the public in 2018, topping all museums worldwide".

Street crossings

"The technology is currently being tested in areas such as street crossings to catch jaywalkers". "At a crossing in Shenzhen, a giant screen displays the faces of pedestrians who dared to jaywalk, unaware they were being watched. But it wasn’t police who caught them — it was street-side surveillance cameras equipped with the latest in facial recognition technology. The miscreants’ mug shots are broadcast on the screen to shame others into compliance".

Detecting face masks

Due to the coronavirus (COVID-19) that emerged in China in 2019, everyone who is outdoors is wearing masks, and it brings a problem to surveillance since normally it is hard to track people’s faces with masks on. Therefore, a Chinese company, Hanwang Technology Ltd (Hanvon), says that they have "come up with technologies that can successfully recognize people even when they are wearing masks". Not only it can identify people’s identities with masks, but it can also detect a temperature over 38 degrees Celsius by connecting to a temperature sensor. The company "used core technology developed over the past 10 years, a sample database of about 6 million unmasked faces and a much smaller database of masked faces to develop the technology". It is now being used in “single channel” recognition, "for example, entrances to office buildings", and “multiple channel” recognition, "for example, multiple surveillance cameras", to detect crime suspects. "It can identify everyone in a crowd of up to 30 people within a second. The recognition rate can reach about 95 percent" when wearing a mask. "The company has about 200 clients in Beijing using the technology, including the police, and expect scores more across 20 provinces to start installing it soon".

Advantages and disadvantages

Advantages

Hamann and Smith have stated that "the use of this technology to enhance public safety will only increase and improve". The advantages of face recognition technology could be summed as follows:

Non-contact

Facial recognition technology plays an important role in our daily life; it is the most direct way of identifying people, without physical contact with machines, the users’ identities could be recognized. Unlike fingerprint that you have to touch the collection equipment with fingers, which is not sanitary, but also easy to cause user's antipathy.

Convenience

Collecting facial recognition data is simple, by using normal cameras, we could complete the face image acquisition process in a few seconds. There is no need for complex equipment.

Friendliness

The method of face recognition is consistent with human habits. Both humans and machines can use face images for recognition. But fingerprint, iris and other methods do not have this feature, a person who has not been specially trained cannot use fingerprint nor iris image to identify other people.

Disadvantages

Privacy

One of the biggest disadvantages/concerns for facial recognition technology is privacy. Taking ZAO app as an example. Its "privacy policy includes a clause which says that its developer gets a 'free, irrevocable, permanent, transferable, and relicense-able' license to all user-generated content, according to Bloomberg. Zao’s privacy policy generated an almost-immediate backlash from users, who bombarded its App Store listing with thousands of negative reviews". It is necessary to further strengthen the security assessment of new technology and new business, take effective measures, and actively prevent the use of its own business platform to implement telecommunication network fraud and other risks. "A survey by a Beijing research institute indicates growing pushback against facial recognition in China. 74% of respondents said they wanted the option to be able to use traditional ID methods over the tech to verify their identity. Worries about the biometric data being hacked or otherwise leaked was the main concern cited by the 6,152 respondents". "At the same time, about 40% of respondents said they have no idea how their facial data has been stored", and "84% of people said they wanted to be able to review the data that facial recognition systems had collected on them and to be able to request that it should be deleted". "Unlike other forms of biological information such as fingerprints or DNA, where consent is usually sought before the data is collected, facial data can be acquired without a person’s knowledge or consent.". Therefore, new regulations are needed to protect individual privacy.

The leak of personal data

The People's Daily "called for an investigation, saying one of its reporters had found face data could be found for sale on the Internet, with a package of 5,000 faces costing just 10 yuan ($1.42)". There are also reports saying that there are some sellers in China who sell face data privately on the Internet, and some post that they can buy 30,000 face photos for 8 Yuan. Some sellers said that they could provide "more channels" of face pictures. Other sellers said that most of the photos they sold came from the moments posted on WeChat without the consent of the other party. Xue Jun, a professor of Law School of Peking University, told the Beijing News that it is necessary to set a threshold for face recognition. “Face information is accompanied by people's lifetime. Once there is a risk of leakage, it is particularly big. Therefore, to carry out the strictest protection, you must first obtain the express consent of the user to collect it. I personally think sometimes it is not enough to get personal consent, and it also needs the authorization of the state to collect sensitive biometric information”.

Future Approaches

Hamann and Smith stated that "progress and improvements in facial recognition are made daily and increased accuracy is foreseeable". "The technology is expected to grow and will create massive revenues in the coming years. Surveillance and security are the major industries that will be intensely influenced by technology. Schools and universities and even healthcare are also planning to implement the facial recognition technology on their premises for better management. Complicated technology used in facial technology is also making its way to the robotics industry". The development of FRT is a necessary trend for society, in the future this technology will be used in a wider range of daily life. At the same time, privacy and personal data should be protected, and the corresponding laws and protective measures should be improved. "Wang Sixin, a professor from the Communication University of China", says that "technology always develops ahead of regulations, you can’t have the law ahead of the technology; the technology will be killed. It is necessary to take some measures to reduce the risks, but facial recognition is a trend. The complaints cannot stop widespread adoption”.