The Potentials of Google Vision API-based Networks to Study Natively Digital Images

Abstract

In this article, we present the potentials of Google Vision API-based networks for studying online images, covering three important modalities as part of a critical visual methodology: the content of the image itself, its specific ‘audiencing’ through web references (or image metadata), and the sites of image circulation. First, we conceptually and technically define different networks built upon computer vision features: image-label, image-web entities, and image-domain. Second, we present a research protocol diagram that illustrates how to build networks of images and respective descriptions or sites of circulation. Third, we discuss the potentialities of computer vision networks as a research device, stressing their data-relational (trans)formations and interpretative specifics. Three different case studies will be introduced as examples. In conclusion, we argue that such a visual methodology requires critical technical practices accounting for the multiple layers of technical mediation involved.

Keywords

• Image analysis
• Networks
• Digital methods
• Visual methodologies
• Computer vision