VSWhat started as an attribute specific to sci-fi movies is now part of everyday life, and the numbers only prove the ubiquitous nature of this biometric flexibility. Driven by security, the fight against crime and by extreme economic competition, the global market for facial recognition software is expected to reach the 11.62 billion dollar mark by 2026. The latter is estimated to be around 3.2 billion euros in 2020, according to data from MarketsandMarkets.
From aviation to automotive, security and finance, this technology has proven itself in recent years. Along with increasing market volumes, facial recognition algorithms are becoming increasingly sophisticated. According to the NIST, the American organization of standards and norms, the accuracy of recognition software has been multiplied by 50 in the space of only 6 years.
In 2014, the average software failure rate was 4% based on 26.6 million faces. It fell to 0.2% at the end of 2018, then to 0.08% at the start of 2020. Like most technologies, facial recognition has its detractors. Fueled by misinformation, many users fear an invasion of privacy and individual freedoms or a potential computer attack. However, facial recognition is among the most secure means of authentication on the web.
Social networks mark the end of confidentiality
“Facial recognition is largely made possible by the desire of Internet users to project their digital portraits on the Web”, according to AVG AntiVirus. Spontaneously, Internet users have given technology everything it needs to open up new horizons. Billions of images of their faces from multiple angles on social networks, creating a solid database for technology to learn and improve thanks to artificial intelligence in particular. And this data is not necessarily stored at Apple, Samsung or even Google.
But there are also a number of databases available for free, although their number of images is limited compared to private or official ones. Among them, Clearview AI, an American start-up that boasts of having recovered 10 billion images from the Internet (social networks, business sites, etc.), including more than 2 billion from Russian social media such as VKontakte . This is why the Ukrainian Ministry of Defense has procured the services of Clearview AI to identify the dead, potential suspects and enemies at checkpoints.
Messy Legislative Patchwork
So far, any technical regulation has proven difficult.
“At the national and international level, discussions are underway for the establishment of standards defining the use of this technology in respect of the privacy of individuals”, specifies the CNDP. The European Union, which had recently considered banning this technology for 5 years, the time to assess its impact, finally decided to let each country choose its own rules on this very ambiguous point. Facebook, which had the facial recognition data of more than a billion users, decided at the end of 2021 to end this system on its platform.
In June 2020, several technology companies such as IBM, Amazon and Microsoft also halted the sale of facial recognition systems to police forces, admitting that they pose problems of discrimination and civil liberties. It was only recently that tech giants Google, Apple, and Microsoft announced an agreement to build a system to authenticate without passwords. The objective, explains Google, is that users can connect to an online service simply by unlocking their smartphone (via their usual method: fingerprint, facial recognition, multi-digit code, etc.). As facial recognition technology improves, its challenges will diminish. In its early days, it was described as frightening, invasive. Today, the dependence of users on it shows all the interest that this technology arouses.