Visão computacional e visualidades contemporâneas: composições do ver e do visível entre a técnica, a ciência e a arte

Abstract

This work approaches, through an image studies perspective, the technological domain of Computer Vision, typically understood by the broad spectrum of algorithmic operations by which a machine is programmed for the automatic interpretation of visual information. It aims at describing the phenomenon of emergence and dissemination of its derived machines, but not from a purely technical point of view, paying particular attention to the modes by which they agency, through their operation, the image and the observer. Supported by the theoretic framework surrounding foucaultian concept of the dispositif, the research understands the studied domain by its concrete instantiations, in forms of regulation and government of the act of seeing and of the visible, therefore deeply traversed by formations of knowledge, institutional demands, observational practices and technical, esthetical and scientific discourses among other elements. In order to make such articulations visible, going from a technical constitution to esthetic, social and political agencies, this works theoretic-analytical course is elaborated through the reunion of multiple objects, selected from technical, scientific and artistic domains, according to their potential for problematizing aspects of this dispositifs agencement. This effort is animated by the methodological gesture of the composition, proposed by Bruno Latour, by which it is sought to build, through connections and approximations, the common ground of the approached phenomenon from the multiplicity of its heterogeneous manifestations. Concerning the modes of agencement of the image and the visible, an alternating movement is described between operations of singling, locating and identifying, in one hand; and generalizing, categorizing and idealizing, on the other. Concerning the agencement of the observer notion taken from Jonathan Crary it is suggested that the visualities of the human and of the machine are mutually constituted through its operation, including processes of taking back to the contemporary modes of seeing the codification of the visible developed within Computer Vision. Through this course it is suggested the paradigmatic value of this dispositif for the comprehension of contemporary dynamics of the production and circulation of the visible, as well as the constitution of new conceptions and practices of observation, even those only indirectly related to its functioning.