Everything you need to know about facial recognition

What is face recognition

Facial recognition is a method of using facial data to confirm an individual's identity. Facial recognition systems can be used to identify people in photos or videos. Facial recognition is a category of biometric security. Other forms of biometrics include voice recognition, fingerprint recognition, and retinal or iris recognition. The technology is primarily used in security and law enforcement, but it has growing applications in other areas as well.

How Face Recognition Works

Many people have learned about facial recognition technology through FaceID, which is used to unlock iPhones. In this application, facial recognition doesn't rely on a vast database of photos to determine an individual's identity -- it simply identifies a person as the sole owner of a device while restricting access to others.

In other applications, the face image acquisition system usually collects face images through snapshots, such as surveillance cameras, smart access control and other equipment. Then the intelligent platform at the back end gathers, processes, stores, applies, manages and shares the relevant data collected at the front end, and combines with the face recognition system to realize a large number of face recognition and apply it in real scenarios such as: face recognition Attendance, face access control, face tracking by the Ministry of Public Security, arresting suspects, etc. Match the faces of people passing by the camera to images of people on the watch list.

Face recognition technology may vary, but in general, it is basically divided into the following steps:

Step 1: Face Detection

Face detection is mainly used for preprocessing of face recognition in practice, that is, to accurately mark the position and size of the face in the image. Find the position of the face from the photo, take the upper left corner of the picture as the coordinate origin, record the coordinates of the upper left corner and the lower right corner of the face frame, and cut out the face part.

Step 2: Face Analysis

In real-world scenarios, the faces captured by the front-end equipment are often not necessarily at the frontal angle, so the face posture in the image needs to be corrected. The key point coordinates of the face are obtained through face key point detection, and the angle of the face is adjusted according to the key point coordinates of the face to align the faces. As shown in the figure, these two faces are completely different faces in the eyes of the computer, so we need to align the faces through some affine transformations.

Use the trained model to automatically mark 68 feature points (landmarks) from the detected face, then look for a standard template in the template library, and use affine change to combine the 68 points with the 68 of the template point alignment.

 

Step 3: Face encoding (feature vector extraction)

A model is trained through a convolutional neural network, and the input model face picture is automatically encoded into a vector with a strong semantic meaning. Training process:

• Enter a photo of a known identity.

• Enter a photo of the same identity.

• Enter a photo of a different identity.

• Repeatedly adjust the parameters so that the photo encodings in Step 1 and Step 2 are as close as possible, and different from the encoding in Step 3 as much as possible.

Step 4: Find matches

Calculate the Euclidean distance between the input image and the vector difference of each picture in the database in turn until the one smaller than our threshold is found. At this point, the face recognition is successful.

Application of face recognition

unlock phone

Many smartphones on the market today are equipped with a face unlock function. This function can better protect the user's privacy data, even if the mobile phone is stolen, the data in the mobile phone can be well protected. Apple claims that the chance of a random face unlocking a phone is about one in a million.

law enforcement

Facial recognition is often used in law enforcement. Police collect mugshots of those arrested and compare them to a facial recognition database. Once photos of those arrested are collected, their photos will be added to the database.

Airport and Border Control

Face recognition equipment is widely equipped at airports. More and more travelers are holding biometric passports, which allow them to avoid long queues and get to the gate faster with automatic e-passport matching. Not only will facial recognition reduce wait times, but it will also allow airports to improve security. The U.S. Department of Homeland Security predicts that by 2023, 97 percent of travelers will use facial recognition technology. In addition, the technology can also improve the security of large-scale events such as the Olympic Games.

looking for missing persons

Facial recognition can be used to find missing persons and victims of human trafficking. Suppose missing personal data is added to the database. Then, whether it is in an airport, a retail store or other public places, as long as the face recognition device recognizes it, law enforcement officers can immediately receive an alarm.

Improve the retail experience

For example, facial recognition devices in stores recognize customers and make purchase suggestions based on their purchase history. "Face payment" can also bring users a convenient payment experience.

banking

Biometric online banking is another benefit of facial recognition. Instead of using a password, customers can authorize transactions by looking at their smartphone or computer. With face recognition, hackers have no passwords to break. If hackers steal your photo database, "inanimate" detection -- a technique used to determine whether the source of a biometric sample is a living person or a fake -- should (in theory) stop them from using it for impersonation purposes.

Marketing and Advertising

Marketers have used facial recognition to enhance the consumer experience. For example, frozen pizza brand DiGiorno used facial recognition technology in its 2017 marketing campaign, which analyzed the expressions of people at DiGiorno-themed parties to gauge people's emotional responses to pizza. Media companies also use facial recognition to test audience responses to movie trailers, characters in TV pilots, and the best positions for TV promotions. Billboards using facial recognition technology - such as London's Piccadilly Circus - mean brands can trigger tailored ads.

health care

Hospitals use facial recognition to aid in patient care. Healthcare providers are testing the use of facial recognition to access patient records, simplify patient registration, detect mood and pain in patients, and even help identify specific genetic disorders. AiCure has developed an app that uses facial recognition to make sure people are taking their medication as prescribed. As biometrics become more affordable, adoption in the healthcare industry is expected to increase.

Track student or worker attendance

Some educational institutions in China use facial recognition to ensure students do not skip classes. Tablets are used to scan students' faces and match them to photos in a database to verify their identities. More broadly, the technology could be used to allow workers to enter and leave the workplace so employers can track attendance.

identify driver

Car companies are experimenting with facial recognition as a replacement for car keys, according to Consumer Reports. The technology will replace keys for entering and starting the car, and remember driver preferences for seat and mirror positions, as well as radio presets.

Monitoring Gambling Addiction

Facial recognition can help bookmakers protect their customers to a higher degree. It can be difficult for workers to monitor people entering and moving around gaming areas, especially in large, crowded spaces such as casinos. Facial recognition technology enables companies to identify those who register as gambling addicts and record their play so staff can advise when to stop. Casinos could face hefty fines if gamblers on the voluntary exclusion list are caught gambling.

Face recognition use case

1. Amazon has previously promoted its cloud-based facial recognition service called Rekognition to law enforcement agencies. However, in a June 2020 blog post, the company announced plans to suspend police use of its technology for a year. The rationale for this is to allow time to activate U.S. federal laws that protect human rights and civil liberties.
2. Apple uses facial recognition to help users quickly unlock their phones, log in to apps, and make purchases.
3. British Airways is offering facial recognition to passengers boarding from the US. Passengers' faces can be scanned by cameras to verify their identities for boarding without showing passports or boarding passes.
4. U.S.-based health insurer Cigna allows Chinese customers to submit health insurance claims that use photos instead of paper signatures to reduce incidents of fraud.
5. Coca-Cola uses facial recognition in a variety of ways around the world. Examples include rewarding customers for recycling at some of its vending machines in China, personalized advertising on its vending machines in Australia, and event marketing in Israel.
6. Facebook started using facial recognition in the US in 2010, when it used a tag suggestion tool to automatically tag people in photos. Since 2019, Facebook has opted in to the feature as part of a drive to be more privacy-conscious. Facebook provides information on how to opt in or out of facial recognition here.
7. Google integrated the technology into Google Photos and used it to categorize images and automatically tag them based on recognized people.
8. MAC Makeup is using facial recognition technology in some of its brick-and-mortar stores, allowing customers to virtually "try on makeup" using an in-store augmented reality mirror.
9. McDonald's already uses facial recognition technology at its Japanese restaurants to assess the quality of customer service provided there, including analyzing whether its employees smile when helping customers.
10. Snapchat was one of the pioneers of facial recognition software: it allowed brands and organizations to create filters that fit users' faces—hence the puppy face and flower crown filters ubiquitous on social media.

Advantages of face recognition

In addition to unlocking your smartphone, facial recognition brings other benefits:

improve security

At the government level, facial recognition could help identify terrorists or other criminals. On a personal level, facial recognition can be used as a security tool to lock down personal devices and personal surveillance cameras.

reduce crime

Facial recognition makes it easier to track burglars, thieves and trespassers. Just knowing that a facial recognition system exists can act as a deterrent, especially for petty crimes. In addition to physical security, cybersecurity also has benefits. Companies could use facial recognition technology instead of passwords to access computers. In theory, the technology cannot be hacked because there is nothing to steal or change, like a password.

faster processing

The process of face recognition only takes one second, which will greatly improve the user experience.

Integrate with other technologies

Most facial recognition solutions are compatible with most security software.

Disadvantages of facial recognition

While some people don't mind being filmed in public or are opposed to using facial recognition when there's an obvious benefit or reason, the technology can spark a strong reaction from others. Some disadvantages or concerns include:

monitor

Some worry that the use of facial recognition and ubiquitous cameras, artificial intelligence and data analytics creates the potential for mass surveillance that could limit individual freedoms. While facial recognition technology allows governments to track criminals, it also allows them to track ordinary and innocent people at all times.

range of errors

Facial recognition data is not without errors, which can lead to people being implicated for crimes they did not commit. For example, a small change in camera angle or a change in appearance (such as a new hairstyle) can cause errors. In 2018, Newsweek reported that Amazon's facial recognition technology misidentified 28 U.S. members of Congress as people arrested for crimes.

invasion of privacy

Ethical and privacy issues are the most contentious ones. The government has been known to store photos of several citizens without their consent. In 2020, the European Commission said it was considering banning facial recognition technology in public places for up to five years to allow time to develop a regulatory framework to prevent privacy and ethical abuse.

massive data storage

Facial recognition software relies on machine learning techniques, which require massive data sets to "learn" to provide accurate results. Such a large data set requires powerful data storage. Small and medium-sized companies may not have enough resources to store the required data.

Biometric information is increasingly captured, stored and analyzed around the world, often by organizations and governments, with a mixed record on cybersecurity. A growing question is how secure is the infrastructure that holds and processes all this data?