Medals and likes: A methodology for big data image dataset analysis of Olympic athletic beauty on Instagram

Abstract

This article seeks to understand how the Cultural Analytics’ methodological approach and computational tools help interpret large image datasets. A set of 87 730 images of 389 Olympic athletes was collected from Instagram and analyzed, featuring a timespan from September 2011 to November 2020. The image set was structured and organized using computer vision processing combined with interactive visualization tools (Google Vision, PixPlot, Image Network Plotter). The analysis, mixing quantitative and qualitative methods, identified patterns represented as image clusters. Regular personal computers served as the hardware platform. Approximately 60 % of the athletes’ posts were related to non-sports topics, highlighting common characteristics of the visual culture disseminated on Instagram, such as selfies, lifestyle, leisure, travel, and food. Images of sports content, considered a central aspect of the research, had a lower frequency of publications featuring topics such as competitions, training, exercises, and sports practices in general. Beyond this result, the study offers a possible technical framework for similar researchers using large image datasets.