Facial Recognition: its operation explained

Facial recognition (in English Facial Recognition) is an algorithmic method that identifies people or confirms their identity according to their biometric facial characteristics. These systems offer an effective verification process and precision greater than conventional methods, but also have certain challenges, in particular with regard to data protection.

What is a facial recognition system?

Facial recognition is a technology allowingidentify and verify the identity of people according to their facial features. Systems are based on the capture and analysis of unique facial biometric characteristics such as the shape of the eyes and nose, which are converted into mathematical models and then compared to a database.

Modern systems can be used to identify people in photos, videos and even in real time. Thanks to these techniques, it is for example possible to compare if a face in two different images belongs to the same person. In addition, facial recognition systems are also able to search for large amounts of visual or video material looking for a specific face.

Facial recognition is a process in several stages which is based on technologies of the fields of computer vision and artificial intelligence. Although there are different systems that vary in their design and functioning, the identification of faces generally follows the following scheme:

• Facial capture: The first step of Detection face consists in locating a face in an image or video, generally using computer vision. This technology is not content to capture face data from the front, but can also capture it in profile.
• Facial analysis: Then the system analyzes the biometric facial characteristics (Face Analysis). Among the main variables are the depth of the orbits, the distance between the eyes, the size of the nose, the shape of the cheekbones and the outline of the lips, the ears and the chin. Most facial recognition systems use 2D images for analysis, as they are easier to compare with public photos and databases.
• Creation of a FACEPRINT: The algorithm converts the facial characteristics captured into a digital signature called « FAPERINT », which mathematically represents the face. As each person has unique facial features, this signature is as unique as a fingerprint.
• Comparison with the database: The system compares the FAPERINT created with a database of known faces and assesses the probability of a correspondence. Thanks to advanced comparison algorithms, it is possible to find correspondence despite variations in lighting conditions, facial expression and shooting angle.

What are the main areas of application of facial recognition systems?

These recognition facial technologies are now used in a variety of applications. The most important areas of application include:

• Smartphones: Many smartphones models allow users to unlock the device thanks to the functions Face ID. According to Apple declarationsthe probability that a random face can unlock an iPhone is less than one in one million.
• Application of the law: In the United States, but also in other countries, the biometric identification method is increasingly used to locate those sought after by the police. Technology even allows agents to take a photo on site with a mobile device and compare it with databases.
• Airports and border controls: An increasing number of travelers has biometric passports, which makes it possible to pass the long queues thanks to the rapid automated passage system at the external borders (Parafe). Facial recognition is also used during major events such as the Olympic Games to strengthen security.
• Banks: Banking applications of many financial institutions provide users with the possibility of authenticating transactions via facial recognition. As it is not necessary to enter a password or a PIN code, cybercriminals do not have the opportunity to steal this information, which improves the security of online banking services.
• Health sector: Systems can rationalize patient recording in hospitals. In addition, this identification method makes it possible to detect emotions and pain in patients.

Facial Recognition: Five practical examples of use

• The e-commerce giant Amazon has developed a cloud -based facial recognition system called « rekognition », which not only allows user -based user verification, but also mood analyzes and video filtering to detect inappropriate content.
• The group Apple Allows its customers to quickly unlock their smartphone thanks to facial recognition. It is also possible to use it to connect to applications and to validate purchases.
• British Airways Allows travelers (in certain airports) to check their identity via facial recognition, thus avoiding the need to present their passport or their boarding pass.
• Coca-Cola Use facial recognition in China to reward customers who recycle bottles and cans. In Australia, the company broadcasts personalized advertisements on its automatic distributors.
• The social media platform Facebook Having used a facial recognition tool in the United States since 2010 to automatically identify people in photos (since 2019, this feature is activated only with user consent).

What role does artificial intelligence play in facial recognition?

Artificial intelligence is essential to the development and functioning of modern biometric identification systems. AI tools allow continuous improvement in technology thanks to automatic learning. These systems use the data provided to adjust their algorithms and thus become more and more effective over time.

Neural networks, and more particularly Neural Networks Convolutional (CNN), are often at the heart of modern facial recognition technologies. These networks analyze the images of faces in several stages, producing facial footprints (hobby) great precision, even under difficult conditions. Their ability to process this data in real time makes it a major asset for sensitive applications, such as monitoring systems and secure access controls.

What are the opportunities and risks linked to the use of facial recognition?

Facial recognition offers considerable potential, especially in the fields of safety and efficiency. Current generation systems allow a Quick and reliable people identificationwhich is useful both for access controls and for the fight against crime and the resolution of crimes. They also contribute to Improve the user experiencefor example as an option of smartphones. Facial recognition allows companies to offer personalized services and thus optimize their processes.

The larger ones concern data protection and privacy. Recognition facial systems to identify and monitor people without their knowledge, there is a risk of abuse by governments, businesses and cybercriminals. In addition, concerns have been expressed by experts concerning the precision of these systems, especially for ethnic minorities, for which the identification error rate is often higher.