Skip to main content
Home
Snurblog — Axel Bruns

Main navigation

  • Home
  • Information
  • Blog
  • Research
  • Publications
  • Presentations
  • Press
  • Creative
  • Search Site

New Approaches to Automated Image Analysis

Snurb — Monday 28 May 2018 22:23
'Big Data' | Social Media | ICA 2018 |

The next speaker at ICA 2018 is Theo Araujo, whose focus is especially on analysing image content from social media. There are a number of API solutions now becoming available for the analysis of such images, including from Google and Microsoft. The project tested such image analysis tools in the context of the visual self-representation of companies discussing their corporate social responsibility.

Such image recognition platforms may not necessarily produce intuitive or meaningful results. The project automatically extracted the most salient labels these tools had provided for the images, and then computationally grouped these into a number of overarching topics; alternative supervised approaches were also tested. These produced mixed results, and a supervised approach turned out to be most successful. Different image analysis APIs also resulted in very different outcomes, of course.

  • 1008 views
INFORMATION
BLOG
RESEARCH
PUBLICATIONS
PRESENTATIONS
PRESS
CREATIVE

Recent Work

Presentations and Talks

Beyond Interaction Networks: An Introduction to Practice Mapping (ACSPRI 2024)

» more

Books, Papers, Articles

Untangling the Furball: A Practice Mapping Approach to the Analysis of Multimodal Interactions in Social Networks (Social Media + Society)

» more

Opinion and Press

Inside the Moral Panic at Australia's 'First of Its Kind' Summit about Kids on Social Media (Crikey)

» more

Creative Work

Brightest before Dawn (CD, 2011)

» more

Lecture Series


Gatewatching and News Curation: The Lecture Series

Bluesky profile

Mastodon profile

Queensland University of Technology (QUT) profile

Google Scholar profile

Mixcloud profile

[Creative Commons Attribution-NonCommercial-ShareAlike 4.0 Licence]

Except where otherwise noted, this work is licensed under a Creative Commons BY-NC-SA 4.0 Licence.