THE HIDDEN FACE OF FACE RECOGNITION

Face recognition is an artificial intelligence system, used to identify or verify the identity of a person starting from one or more images that depict it.

Face recognition occurs through digital image processing techniques and is based on pattern recognition, whereas the pattern is a human face, neglecting everything that is not part of the face and therefore only becomes part of the background.

The practical applications are interesting and derive essentially from the possibility of recognizing the person in front of a camera, video camera, webcam or another digital photographic device. We use it to identify the users of smartphones and computers, in the control of people’s access to certain areas subject to some limitation, in the search and identification of a certain person, if present in a special photographic database.

In these areas, the big company in American technology has been working hard for years.

Amazon from 2016 offers its corporate customers “Amazon Rekognition”. The product uses deep learning, which is an artificial intelligence technique for pattern research. In the case of some specific experiments, carried out in the Florida and Oregon police departments, the patterns were the faces of people who were confronted in real-time with a database of suspicious individuals.

As we can imagine, these technologies allow us to quickly overcome the limit, always very subtle, of the violation of privacy and the civil and human rights of people.

In this regard, the shareholders and investors of Amazon have urged the company to suspend sales of face recognition software to government agencies and to make an independent assessment to verify whether this technology does not violate the facts or potentially the aforementioned rights.

Both for technological reasons and for these solicitations, Amazon does not actually use face recognition in its futuristic stores without cash desk  “Amazon Go”, where customers are identified by a log in a specific application and not by their biometric data.

Google has got a series of specialized facial recognition products, part of the “Pixel” family. Notwithstanding the main purpose, which is to control access to the smartphone of only authorized users and support a series of photographic features, it seems that these systems have highlighted some uncertainty in the recognition of some facial shapes and skin tones, which can be associated above all with colored people. In addition, the New York Daily News pointed out in an investigation that, to increase the statistical data of deep learning and to overcome this anomaly, their suppliers used practices that were on the verge of fairness to steal images of homeless and disadvantaged people in the suburbs of some big American cities. For these situations, Google has subsequently limited recruitment to actual volunteers, increasing transparency with more information on the purposes of the activity carried out.

Microsoft, which has a series of products including “Viso” and “Windows Hello”, through its CEO in a recent economic forum, has asked the US government more attention on privacy, hoping for the introduction of new regulation, particularly regarding to face recognition technology. In support of the concerns expressed, it has also recently canceled MS Celeb, a database with about 10 million images belonging to about 100 thousand public figures, which was used to refine its recognition algorithms.

In China, the government has for years shown the objective of achieving the technological supremacy of the world and the scope of face recognition is certainly no exception. Huawei, Hikvision, SenseTime, and Megvii are Chinese market leaders in digital image processing. A recent New York Times investigation revealed that face recognition appears to have been used to monitor, in some areas of the country, an ethnic minority, considered hostile to the Chinese government, through 500 thousand face scans per month. Perhaps, for this reason, the government of Washington has included these Chinese firms in the Blacklist of suppliers banned from doing business in the U.S.

In England, the recent experimentation of a crime prevention system, based on face recognition, performed by the Metropolitan Police in some areas of the London capital (Soho, Piccadilly Circus, Leicester Square, and Westfield) has been echoed. The results, however, have been disappointing, since an error margin of 80% has been highlighted, still too high to engage the police in control interventions based on surveys that are often incorrect. The same results had already been reached in similar experiments, but carried out in the US, in San Francisco, Somerville as well as Oakland.

Regardless of the maturity and reliability of the technology used, in this context, we would like to underline the seriousness of an unscrupulous and uncontrolled use of data deriving from face recognition, which can have a serious impact on the civil and human rights of the person (let us think of identity theft, economic fraud, uncontrolled disclosure of information covered by professional secrecy or relating to the religious, political, ethnic or sexual sphere and the prejudice to the reputation that all this could entail).

Fortunately, since May 2018, Europe has a cutting-edge restrictive law: the General Data Protection Regulation (GDPR). Thanks to this law, in none of the 27 member states of the European Union and therefore not even on the Italian territory, it is possible to collect biometric data without the explicit consent of the data subject. Furthermore, this data, considered as special categories of personal data, cannot be used except in a few specific cases: in the context of judicial proceedings, in the public health sector for health security purposes or for reasons of public interest.

ACTION ICT is a young, dynamic and innovative IT company. It operates, both nationally and internationally, offering professional skills and design solutions in the ICT field to medium and large businesses. Our highly skilled know-how is housed in three expertise specific departments: ACTION DATA (Big Data and Analytics), ACTION APP (Web & Mobile Application) and ACTION IOT (Artificial Intelligence, Internet of Things and Robotics).