Technical image forensics as a tool for the content analysis of fitspiration social media posts – Highlighting gender and diversity aspects of online body depictions

Aim and Research Question(s)

The trending social media tag of #fitspiration, a combination of fitness and inspiration, involves users of social media, showing of their physically toned bodies in a fitness context. This thesis aims to provide an approach towards understanding as to what can be done to make #fitspiration image manipulation easily and more reliably recognizable as fake to the untrained eye. Furthermore the question arises whether technical image manipulation in #fitspiration posts can be reliably detected and if so, with which frequency does it appear and in which forms does it occur? (i.e., pixel cloning, smoothing etc.)

Background

With the underlying intent to promote a healthy lifestyle and to showcase idealizations of the human body, fitspiration imagery could influence the bodily self-awareness and gender identity of its consumers, both negatively and positively [1]. By giving users the tools to actively engage with the digital image manipulation taking place, a healthier outlook on bodily health, beauty standards and diversity could potentially be developed [2].

Methods

To highlight which forms of image manipulation occur amongst gender and diversity of posters, one hundred (100) #fitspiration posts where manually sampled from the platform Instagram over a time span of four weeks and analyzed with technical and visual processes of image manipulation. To gain insight into whether technical image manipulation of #fitspiration posts is detectable by lay users of social media with and without aid, a randomized controlled pilot trial with parallel groups, at a high school in Vienna was carried out (N104). #fitspiration top post, Retrieved 10/09/2022 with 10,352 likes

Results and Discussion

The compression level present within the Instagram compression algorithm made the technical detection of altered areas nigh impossible. A tendency towards users with feminine features employing the use of image manipulation could be observed. Visually detectable image manipulation was most common in the form of warping. User testing suggests that the visual and technical analysis carried out within this work, is in line with what lay users would also deem as fake in 50 percent of the cases.

Conclusion

#Fitspiration favors female features (N62) over male ones (N38). Analysis suggests a focus on muscle and body mass in males and a more thinness centered approach towards #fitspiration in females. Image warping was determined to have been the culprit in 19 out of 20 cases deemed highly likely to be faked. Users with female features did so 15 times, with a focus on the hip and buttock area. Further research suggests neural networks as a future tool for the detection of image manipulation.

References

[1] Norton, M. (2017). Fitspiration: Social Media’s Fitness Culture and its Effect on Body Image. [2] Robinson, L., Prichard, I., Nikolaidis, A., Drummond, C., Drummond, M., & Tiggemann, M. (2017). Idealised media images: The effect of fitspiration imagery on body satisfaction and exercise behaviour.