Significant progress has been made in recent years in the field of image processing, particularly in facial recognition. The number of deployments and experiments of this type of system is rapidly increasing. Most applications are motivated either by security or by commercial considerations. However, there are different opinions regarding the use of these systems, particularly in the public space. Considering the lack of consensus on a technology that can have a significant impact on society, many organizations have alerted public opinion and called for a public debate on this topic. We believe that such a debate is indeed necessary. However, for such a debate to be productive, it is essential to ensure that arguments can be expressed and confronted in a rigorous way. In particular, it is critical to avoid, as much as possible, preconceptions and to distinguish established facts from assumptions or opinions. The purpose of this document is precisely to set the terms of the debate on a solid basis. Our aim is not to take a position on facial recognition in general or to provide an exhaustive review of its applications, but rather to propose some elements of methodology for the analysis of its impacts, illustrated by some examples. The importance of conducting impact assessments of facial recognition technologies is underlined by institutions such as the European Commission and the European Union Agency for Fundamental Rights (FRA) but there is still a lack of methodology to put them into practice. As facial recognition applications are very different, it is necessary to precisely analyse the potential impacts of each system, taking into account all its features and the context of its deployment. However, this case-by-case analysis should not overlook the more "systemic" risk related to a potential generalization of the reliance on facial recognition in our societies. This global risk must also be analysed and debated because it may justify, according to some people, a total or partial ban of facial recognition.