An exploratory study on joint analysis of visual classification in narrow domains and the discriminative power of tags

  • Authors:

  • Oge Marques;Mathias Lux

  • Affiliations:

  • Florida Atlantic University, Boca Raton, FL, USA;Klagenfurt University, Klagenfurt, Austria

  • Venue:
  • MS '08 Proceedings of the 2nd ACM workshop on Multimedia semantics
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

The popularity of social media sharing sites such as Flickr has driven a significant amount of research on the analysis of information contained in the tags used to annotate images. Many of such tags are not useful to describe the contents of an image and are often labeled as not descriptive or even noisy. In this work we focus on the descriptiveness of a tag in an exploratory way, within a relatively narrow domain, and with the help of a visual classifier. Preliminary experimental results demonstrate the possibility to infer descriptiveness of tags from a joint analysis of tag entropy calculations and the results of an automated visual classifier with a limited number of classes without taking tag content or tag co-occurrence into account. We postulate that these experiments can be extended and improved toward a working solution that might answer the question: Given a semantic category, which tags would you use for searching an image from that category?