Request demo
en
Site language:
Biometric software product with robust user authentication for unlocking a device or gaining access to operating systems or apps
A biometric software product for increasing the level of security at the facility during video surveillance
A biometric software product for biometric identity verification in access control and management systems
A biometric software product for displaying personalized media content
A biometric software product with facial recognition for reliable identity check
A biometric software product with facial recognition that expands the electronic queue systems with personalized services for visitors
A software product that provides simple and reliable working time and attendance by identifying faces using tablets, IP cameras, or terminals
A biometric software product with facial recognition that provides enhanced communication with clients
A biometric software product that provides a reliable and quick check of the gym clients access right without employee’s participation
A new level of work with visitors and employees of Business centers opened with the help of biometric products
Biometrics for convenient service to citizens, including remote monitoring of the quality of personnel work
Biometric monitoring of working hours and additional security tools for industrial facilities
Modern methods of biometric analytics for safe operation of sports facilities
Convenient and secure transport solutions based on the digital identity of the passenger
Biometric solutions for a new level of security and interaction with visitors
Biometric video Analytics for targeted marketing and personnel control in distributed networks
Biometric products for proctoring and video surveillance systems in educational institutions
Keyless biometric access to rooms, targeted approach to each client and information about the time of work for employees
Necessary tool for the security and competitiveness of a modern Bank
Improving the level of security, speed of investigations and timely prevention of illegal acts in the urban public space
Customer-oriented solutions, acceleration of the work process of the registry area, control of the staff of the entire institution
RecFaces makes facial biometrics simple and applicable. We provide a wide range of ready-made biometric solutions for businesses to upgrade their security and technological efficiency.
We are ready for cooperation and sales through the partner network. To get advice on your project, please contact us by e-mail sales@recfaces.com
It is our principal and invaluable resource. Talented and energetic people of our team unite the like-minded ones which helps to expand expertise and company’s growth.
Join us!
We are always happy to answer all your questions. Contact us in any way convenient for you.
We share our long-term expertise
in the development of biometric software
Comprehensive information for your projects.
Just theory, practice and statistics
Discover our latest news and updates
on facial recognition technology
Find out more about
RecFaces company here

Twins, plastic surgery and cheating makeup

Twins, plastic surgery and cheating makeup — RecFaces

Facial recognition technology penetrates deeper and deeper our daily life every year.

Unlocking gadgets, paying by face in public transport and stores, opening bank accounts remotely and even passing passport control at the airport — the number of application fields of facial biometrics is constantly growing. Yet facial recognition remains surrounded by myths and delusions. In this article, we collected the top popular and exciting questions about facial biometrics.

Contents

How does the facial recognition work?
Does biometrics can distinguish between twins?
Will a biometric solution recognize a person, if he or she has changed with age?
Will biometrics recognize a person if he or she start wearing a beard or glasses?
And if you apply special makeup?
Will biometrics recognize a person after plastic surgery?
And how does biometrics distinguish representatives of different nationalities?
Is it possible to make a biometric passport for a small child?
Is the task of biometrics to track people?
How easy is it to hack biometrics? Is it possible to deceive the system using photos or videos that a person shares on social networks?
Can the theft of biometric data harm the user somehow?

How does the facial recognition work?

The main way to recognize faces today is neural networks. Their advantage lies in the ability not only to analyze information, but also to train in process and remember new data. The recognition process itself consists of several stages. First, you need to get an image of the person you want to identify. The system captures the picture from the CCTV camera, detects the face and cuts it out, aligns and sorts it. After that, anthropomorphic points are placed on the image of a person. They are located in key areas of the face: along the eye contour and between eyes, on the forehead, chin, in the area of the lips, ears, etc. The total number of anthropomorphic points can reach a few dozen. The neural network measures and sums up the distance between them, which allows you to calculate the geometry of the face. The next step is to convert the dot picture into a digital code or «face print». The image turns into a unique set of numbers. At the last stage, the obtained code is compared with the codes of people whose data is stored in the database of biometric profiles. A match means the identification is passed. At first glance, this process seems to be a very long one, but in reality, the neural network works incredibly quickly. Thus, recognition algorithms installed in RecFaces biometric products take less than 1 second to identify one person.

Does biometrics can distinguish between twins?

Biometric algorithms distinguish twins better than the human eye, but the probability of an algorithm failure when identifying them is still slightly higher. There are two types of errors in biometrics: the first and second kind. The first is a false failure when the system does not recognize a person, although his data is in the database. The second is false access, when the system takes a stranger for a person from database. Error probabilities can be lowered by training facial recognition algorithms or setting up the biometric product. Thus, it is possible to raise or lower the similarity score in the system. In the case of twins, choosing the optimal settings is more difficult: both the too high and too low similarity scores mean the probability that the system will either confuse the twins, or mistake them for one person. If the similarity score is too low, the system can stop distinguishing them, and if it is too high, it can take one for another. However, the probability of error increases only by hundredths of a percent, so the problem is not crucial. Besides, additional identification factors can be used for areas where a 100% security guarantee is required. For example, passwords or RFID cards. Another option is to combine facial recognition with fingerprint or iris identification. These signs are different in twins. Solving the problem of twin identification is one of the important directions in the development of facial recognition technology. Google has invested more than $2 billion in research. Different countries are engaged in The search for a solution to the problem. So, in July 2022, a group of mathematicians presented their method of recognizing twins at a scientific conference in the UAE. The essence of the development is that in a photo from a database or video stream, brightness and contrast are corrected, colors are aligned along the channels of the RGB color model and passed through the Sobel operator, making the picture black and white. According to the developers, the method allows 99% accuracy when recognizing not only twins, but also the same person at different ages.

Will a biometric solution recognize a person, if he or she has changed with age?

Our face changes with age, which leads to the so-called obsolescence of the biometric template. So, biometric data must be updated periodically to maintain its relevance. Firstly, age-related changes in our appearance do not always affect the location of anthropomorphic points. Secondly, the appearance changes very slowly. This process is almost invisible for us and our loved ones. But the changes that occurred will immediately notice a former classmate, who has not seen us for 15 years. The same way it works with biometric algorithms. Let's say the business center introduces biometric ACS at the entrance and makes photo of a person for a profile base. After several years, his face gradually changes. At some point, he ceases to be similar to his photo from the database, but the device at the checkpoint continues to allow him to enter. Why does this happen? It is all about the ability of modern neural networks to self-train, which we mentioned at the beginning of the article. Every day the program remembers even the smallest transformations of appearance and updates our image. Almost like members of our family. It's another case if you initially try to put your 15-year-old photo in the profile. Then the neural network will already play the role of a former classmate: it can recognize you, or maybe not.

Will biometrics recognize a person if he or she start wearing a beard or glasses?

Hairstyle, beard, mustache and glasses do not affect the algorithm efficiency. A person's digital code expressing facial characteristics depends on anthropomorphic points. They remain the same regardless of the use of glasses or the change of style. Moreover, during a pandemic, modern neural networks have perfectly learned to distinguish between people in medical masks. For that purpose, the algorithms were reoriented to read anthropomorphic points located on the eye area, and neural networks had to be re-trained on databases with photos of people in masks. The first company that announced the solution of the problem was the Chinese Minivision in February, 2020. At the beginning of the pandemic, engineers took up additional training of algorithms in photos of employees, acquaintances and relatives. It took less than two months to finalize the program.

And if you apply special makeup?

Here it all depends on the goals a person pursues. If the goal is to pretend being someone else, then this is impossible. Let us explain. It is unrealistic to copy the geometry of another person's face either using the most masterful makeup. If the main goal is to remain unrecognized for the system, then the chances are slightly higher. People have been trying to cheat facial biometrics since it appeared. Catchy makeup, stickers and fancy jewelry on the face, chameleon glasses, medical masks with a face image are often used. For example, in 2020, members of the British association of artists Dazzle as a sign of protest against the use of biometrics in urban video surveillance systems walked around London in special makeup. They painted bright contrasting triangles and stripes on their faces. According to activists, they managed to become unrecognizable for the system for unlocking their smartphones using makeup. However, large biometric systems used in banks, office centers and street video surveillance networks today use more accurate algorithms than in phones. Deceiving them with makeup or glasses is much more difficult.

Will biometrics recognize a person after plastic surgery?

Significant changes in appearance require updating the biometric template of a person. This includes eye loss, changing the shape of the eyes, rhinoplasty, facelift and other major surgical procedures affecting facial geometry, but scars, lip augmentation and Botox do not affect facial recognition.

And how does biometrics distinguish representatives of different nationalities?

The principle of facial recognition algorithms operation remains unchanged regardless of race. Moreover, as in case of twins, the neural network distinguishes faces much better than the average person. However, it can depend on how the neural network was trained. A program developed in China that was trained by using mainly databases with photos of Asian people, can recognize Caucasian faces worse, and vice versa. So, in 2020, during the protests of the BLM movement in the United States, they started talking about discrimination in biometrics. Developers of algorithms were reproached for the fact that programs too often fail to identify African-Americans and mistake them for criminals. In response to the accusations, the International Association of Biometrics and Identification (IBIA) checked the work of a number of large algorithms. The report on the results of the study was intended to refute allegations of biometric intolerance. Experts stressed that the accuracy of recognition does not depend on a person's race or bias, but on the quality and features of the algorithm itself. Major world developers use databases with photos of people of different races, when training neural networks. Despite the fact that the probability of error still persists, it is not always related to the demographic issue. So, some algorithms are better at recognizing women than men.

Is it possible to make a biometric passport for a small child?

The introduction of biometric passports in the world began in the first half of the 2000s. They are distinguished from ordinary documents by the presence of a chip where a person's digital photo and other biometric information are recorded. These may be fingerprints, iris patterns, or other data. Today, biometric passports are used in more than 150 countries. New generation passports allow you to use biometrics actively in the travel and air service industry: for passing pre-flight formalities and passport control. However, the weak point in the introduction of biometric passports remains the possibility of issuing such documents for children. World experience gives different options for solving this problem. So, in Germany, citizens aged 1 to 12 years are allowed to issue not a biometric passport of a European model, but a simple passport of a child. Such a document is suitable for movements within the European Union, but in most countries outside the EU it will not be accepted. In Spain, biometric passports are limited to two years for children under five.

Is the task of biometrics to track people?

Facial recognition is an important tool in the work of law enforcement agencies. Police and intelligence agencies use smart CCTV to investigate crimes, detain lawbreakers, tackle terror attacks and search for missing people. However, biometrics is not only about maintaining order, it is about convenience in everyday life. Pay in stores and withdraw cash from ATMs, protect your phone in case of theft, secure online banking — facial recognition makes our lives easier, safer and more comfortable. RecFaces stands for such good biometrics. As for the prospects for the further introduction of biometrics, everything will depend on the legal framework. In this sense, biometrics can be compared to weapons: it all depends on how and when to use them. For the further development and application of face recognition technology, it is important for the state to develop clear and transparent legal norms, having previously put forward the problems of data collection and storage for general discussion.

How easy is it to hack biometrics? Is it possible to deceive the system using photos or videos that a person shares on social networks?

Whether it is possible to deceive biometrics by impersonating someone else when identifying — It is a question that has troubled many people. Special masks, photos and deepfake videos are used to attack biometric systems. The reason for special concerns is most often video and photo content, because both attackers can easily be found on social networks. However, it is important to understand that a photo of very high quality and resolution is required to create a deepfake. It is impossible to deceive biometric protection with a simple picture from social networks. And in general, hacking biometrics today is an extremely time-consuming task. The fight against compromise is a key vector in the development of modern biometric technologies. There are several approaches to defense. The first is the analysis of the image in different spectra: ordinary, infrared, 3D infrared. When checking, the system compares the balance of a full-color picture and an infrared image. The second is the algorithms for checking liveness. Their main task is to make sure that there is a real person in front of the camera. Most often liveness uses online verification. A person is offered to perform a number of actions: blink, turn his head, smile. The system considers not only the fact of commands execution, but also the naturalness, as well as the homogeneity of the surrounding textures during movement. Some algorithms apply texture analysis: they evaluate glare on the skin, light areas and shadows on the face. Another option is frame analysis. It allows you to make sure that the face is against a natural background. A photo of a person on a smartphone screen will not pass such a check: the system recognizes the presence of a gadget frame.

Can the theft of biometric data harm the user somehow?

The fear of people about theft or loss of biometric data, unfortunately, is quite common. Due to the lack of awareness of the technology principles, people often tend to exaggerate greatly the consequences of such incidents. In modern biometric systems, all information is stored encrypted in different databases. Photos of people, biometric templates and personal information are converted into a set of numbers. The biometric template is not a key, a bank card or a phone password. Without the correct algorithm, it is impossible to restore the template, and it is actually impossible to guess the right algorithm. Moreover, biometric data makes a person less vulnerable to scammers. Biometrics is the best weapon against social engineering, the methods of which are widely used by attackers. PIN codes and one-time passwords for banking operations can be easily lured out by talking to a person or posing as an employee of a financial institution. These techniques are hopelessly outdated compared to biometrics: if victims of such frauds had biometric identification set up, they would never lose their money. After all, unlike bank card data, it is impossible to “lure” his face or fingers from a person. This means that access to confidential information and the risk of compromise tends to zero. Biometrics is about security and convenience. Like all new technologies, facial recognition technology raises many questions and prejudices. However, every year the benefits of facial biometrics for an ordinary person become more obvious. Popularization of knowledge about biometrics and its benefits is one of the RecFaces missions. You can learn more about biometric solutions based on facial recognition technology and the application of biometrics in different areas of life in our blog and on free RecFaces webinars. What misconceptions and myths about biometrics do you know?

Here you can rate our article
Here you can rate our article
Thanks!

Subscribe to our newsletter