TY - JOUR

T1 - A computer-assisted analysis of image representations of obesity

T2 - Comparing UK news content with the World Obesity Federation image bank

AU - Collins, Luke

AU - Baker, Paul

PY - 2024/2/21

Y1 - 2024/2/21

N2 - Visual representations in news media contribute to the social construction of health concerns such as obesity. Media representations of obesity, however, have been shown to be stigmatizing, focusing on disembodied abdomens or depicting people in ill-fitting clothing, for example. In addition, analyses of image representations have typically focused on small datasets and relied upon inductive thematic coding. This article presents the application of Google Cloud Vision – an automated image annotation tool – to a collection of images collected from articles on obesity in the UK press. In addition, the World Obesity Federation (WOF) has produced an image bank designed to support journalists in offering respectable depictions of people with obesity, alongside its general guidelines for media representations. The authors compare the images from the news with the WOF Image Bank, on the basis of the tags generated by Google Cloud Vision. They use corpus methods to highlight frequently occurring tags that demonstrate similarities and differences between what is represented in the news and what is provided by the WOF. They observe only minimal consistencies between the datasets in that, while both image collections often depict people with obesity in relation to food, the WOF Image Bank provides a greater variety of activities that include food purchase and preparation. In the news, the analysis finds a greater occurrence of ‘body positive’ representations but also a continuation of the focus on abdomens and pinched skin. In the WOF Image Bank, the authors observe more ‘body neutral’ representations, with people engaged in a greater range of recreational activities and socializing with others. The article reflects on the utility of Vision as an automatic tool for capturing salient elements of visual representations in images.

AB - Visual representations in news media contribute to the social construction of health concerns such as obesity. Media representations of obesity, however, have been shown to be stigmatizing, focusing on disembodied abdomens or depicting people in ill-fitting clothing, for example. In addition, analyses of image representations have typically focused on small datasets and relied upon inductive thematic coding. This article presents the application of Google Cloud Vision – an automated image annotation tool – to a collection of images collected from articles on obesity in the UK press. In addition, the World Obesity Federation (WOF) has produced an image bank designed to support journalists in offering respectable depictions of people with obesity, alongside its general guidelines for media representations. The authors compare the images from the news with the WOF Image Bank, on the basis of the tags generated by Google Cloud Vision. They use corpus methods to highlight frequently occurring tags that demonstrate similarities and differences between what is represented in the news and what is provided by the WOF. They observe only minimal consistencies between the datasets in that, while both image collections often depict people with obesity in relation to food, the WOF Image Bank provides a greater variety of activities that include food purchase and preparation. In the news, the analysis finds a greater occurrence of ‘body positive’ representations but also a continuation of the focus on abdomens and pinched skin. In the WOF Image Bank, the authors observe more ‘body neutral’ representations, with people engaged in a greater range of recreational activities and socializing with others. The article reflects on the utility of Vision as an automatic tool for capturing salient elements of visual representations in images.

KW - corpus linguistics

KW - image annotation

KW - media guidelines

KW - news

KW - obesity

U2 - 10.1177/14703572231209481

DO - 10.1177/14703572231209481

M3 - Journal article

JO - Visual Communication

JF - Visual Communication

SN - 1470-3572

ER -