Biometrics technology is used to identify and authenticate a person using a set of recognizable data that is unique to them. Face recognition technology (aka computer vision technology) can be used to recognize or verify a person’s identity in just a few seconds based on their facial features: the distance between the eyes, lip contours, chin, etc. When it comes to facial biometrics, a 2D or 3D sensor scans a person’s face and then converts the information into digital data. After that, a special algorithm compares the scanned image with those stored in the database loaded into the system.
Let’s see what’s so revolutionary about facial recognition technology and how it affects different business domains.
Security and law enforcement
Against the backdrop of the increased cyber crimes and terrorism, security and law enforcement are at the forefront of implementing the face recognition technology. The advantages of using the technology are obvious: timely crime detection and prevention.
Face recognition is often used alongside other AI technologies such as fingerprint recognition. The technology is used extensively for a variety of security purposes, from issuing ID documents to customs control at the airports, etc.
Trending AI Articles:
1. Tutorial: Stereo 3D reconstruction with openCV using an iPhone camera
2. Machine Learning using Logistic Regression in Python with Code
3. Bursting the Jargon bubbles — Deep Learning
4. How Can We Improve the Quality of Our Data?
In 2017, Gemalto, a company specialized in digital security solutions, introduced a new control system for the Charles de Gaulle airport in Paris based on facial recognition technology. In particular, this solution was developed to move from fingerprint scanning and recognition to face recognition gradually.
The police often use biometric facial identification, and its application is strictly controlled in Europe. In 2016, the “man in hat” accused of committing a series of terrorist acts in Brussels was identified precisely through an integrated FBI facial recognition system. In 2017, the South Wales Police also used this technology during the UEFA Champions League final to identify troublemakers. And the Chinese law enforcement agencies use special glasses based on augmented reality and facial recognition technologies that can promptly identify intruders by comparing a person to a database of criminals.
Combined with airborne cameras, drones offer an interesting solution for facial recognition use cases during mass events. According to the 2018 Keesing Journal of Documents and Identity, some drones can carry cameras as heavy as 10 kg and more, which can recognize a potential criminal from a distance of 800 meters from a 100-meter altitude. Moreover, communication with ground control cannot be intercepted.
Retail & FinTech
In 2017, one of Walmart’s largest retail chains introduced its own solution based on face recognition technology. The system can determine the mood of each customer while shopping, as well as measure customer satisfaction right after visiting a supermarket.
Thus, if the system finds a customer with an unhappy or disgruntled face, the store employee will be notified. Walmart believes that this innovation will help improve its customer service and increase customer satisfaction rate. After all, the employee will be able to respond quickly to customer dissatisfaction and help them solve any problem. The system will also help analyze customer behavior over a specified period.
Also, this innovative service can compare customer emotions with what they buy and how much they spend. This helps spot the change in customer habits depending on the amount of money spent on shopping.
It is also worth mentioning the fact that in 2015, American retailers were ready to use face recognition technology in their stores to prevent theft. The irony is that Walmart was one to have rejected the idea, as the system proved highly unprofitable at that time. Besides, many human rights activists opposed the introduction of such a system in supermarkets, as it violates user privacy. Chinese retailers, on the other hand, believe that face recognition technology is safer than scanning fingerprints, retina or using traditional passwords. Therefore, in 2018, the country’s supermarkets started testing the technology at the cash desks.
As a customer, you can now complete check-out at the autonomous cash desks. First, you need to scan barcodes of all products in the basket, and then the camera built into the terminal dashboard will recognize whether your face matches the one assigned to your online account. To complete the purchase, you need to enter your mobile phone number, which must also be linked to the customer’s online account. Experts are confident that face recognition technology will be able to eliminate the sale of cigarettes and alcohol to minors in supermarkets and kiosks. Appropriate software will be introduced at self-service cash desks at the UK supermarket. A pilot project was presented last year, and in 2019, biometrics is expected to be used in supermarkets widely across the United Kingdom.
Speaking about computer vision in FinTech, American Express Co. is ahead of the curve. According to the Wall Street Journal, the company is experimenting actively with facial recognition and wearable technology in search of new capabilities for some of its mobile apps, including solutions aimed at customers traditionally underserved by financial services companies.
Some time ago, American Express launched its own innovative lab that’s getting support from senior leadership. The aim of this lab is to take advantage of the emerging tech in order to offer better and safer services to customers and stakeholders.
American Express Enterprise Growth has been testing ways to capture and authenticate face images accurately and quickly on a mobile device. A customer could use it when conducting sensitive transactions via mobile devices such as money transfers. However, the product isn’t market-ready yet, as the researchers have stymied into bottlenecks and are now working to improve the accuracy of facial recognition.
Significant advances in the use of facial recognition technology can be traced back to medicine and healthcare. For example, through careful face examination and analysis, it is possible to monitor the patient’s use of medications more accurately, find genetic diseases such as DiGeorge’s syndrome with a 96.6% precision, and control the anesthesia procedure during surgery.
The popular fast food restaurant chain, KFC, in partnership with financial services company Ant Financial, a division of Alibaba, launched an unusual payment service in 2017 called Smile to Pay. The novelty appeared in one of the fast food establishments in Hangzhou. The service was first introduced in 2015 by Jack Ma, Founder of Alibaba, who presented it at the international CEBIT exhibition.
To buy a dish from the menu, you only need to take two steps: 1) enter the phone number that’s connected to the Alipay wallet, and 2) smile at the camera built into the biometric terminal. Then follows the process of confirming the identity of the wallet owner, after which the payment is successfully carried out. Smiling is required to enable the system to understand if there is a living person in front of it and not just a picture. The process of recognition occurs instantly in 2–3 seconds.
As you can see in the video above, there’s no way to fool the system: even if you change your hair, hair color, and makeup, the service determines the unique individual features of a person. Last year, face recognition technology from Alibaba was tested in two Marriott hotels in Hangzhou, as well as in Sanya County (Hainan Island). To get the room keys, the guest has to go to a special self-service center, where they have to scan their passport, enter contact details and take a picture. The system then compares the photo in the ID with the guest’s face. If the system confirms that it is the same person, then he or she receives the key-card for the room. Introducing such a system in hotels will reduce the check-in time from 3 minutes to only one minute.
Besides Marriott, the face recognition technology is planned to be used by a Chinese Airbnb analog Xiaozhu. In particular, the company is going to replace regular locks with “smart” locks equipped with the latest face recognition system. With this system put in place, the procedure of short-term rental housing will be safer and comfier. Face recognition technology will help landlords rent out their apartments remotely, i.e., they won’t have to come every time to give the keys to the guests who have booked the apartment.
Is there a way to fool face recognition technology?
Despite many advantages of biometric identification systems, there are also many opponents of these technologies claiming that there’s no system out there that can’t be cracked and compromised.
Let’s review some of the technology failure cases.
German artist Adam Harvey demonstrated how facial recognition systems and CCTV cameras can be deceived by changing one’s hair and makeup. This is what his project called CV Dazzle is all about. Harvey argues that facial recognition algorithms focus mainly on the distribution of light and shadows on individual areas of the face: mostly on the cheekbones, nose bridge and chin. Thus, if you disguise them with a different makeup and hairstyles, you can easily confuse the cameras.
At the end of 2017, cybersecurity experts of one Vietnamese company used a 3D-printed mask to hack into a Face ID function of the new iPhone X. They could prove that the system factually uses only parts of a person’s face for identification, so it can easily be confused and cheated. At about the same time, cybersecurity experts from a German company hacked into the Windows Hello function, which can be used to access Windows 10 through facial recognition. ferentThis was done using a photo printed on a regular printer.
An article in Forbes says researchers at the University of Toronto have developed an algorithm to disrupt facial recognition. In short, the user can use a filter that changes certain pixels in the image. These changes are invisible to the human eye and are very confusing for facial recognition systems.
A February 2018 study by the Massachusetts Institute of Technology found that proprietary facial recognition systems of tech giants such as Microsoft, IBM, and Megvii often make errors. Thus, none of the systems could correctly identify women of the Negroid race. In May 2018, it was reported that the American Internet retailer Amazon began to actively promote its own cloud-based face recognition service called Rekognition to law enforcement agencies. The software can recognize about 100 people in one image, comparing their faces with tens of millions of people in different databases.
Speaking about the developments around face recognition technology, Facebook launched its DeepFace program back in 2014. This service could determine whether the two photographed faces belonged to the same person with an accuracy of 97.25%. However, when people took the test, they gave correct answers in 97.53% of cases, which is more than 12% better than the FBI’s Next Generation Identification system.
In conclusion, while face recognition technology is still in its infancy and leaves much to be desired in terms of accuracy and reliability, there’s no way any modern industry can do without it. As we already live in a heavily digitized world, the cybercrime rates increase from year to year, driving both innovative startups and established brands to increase their face recognition R&D investments and collaborate with AI labs and software development teams all over the world in search of new tools that will help take face recognition to the next maturity level.