Discovering Pictorial Brand Associations from Large-Scale Online Image Data

Gunhee Kim, Eric P. Xing; Proceedings of the IEEE International Conference on Computer Vision (ICCV) Workshops, 2013, pp. 404-411

In this paper, we study an approach for discovering brand associations by leveraging large-scale online photo collections contributed by the general public. Brand Associations, one of central concepts in marketing, describe customers' top-of-mind attitudes or feelings toward a brand. (e.g. what comes to mind when you think of Burberry?) Traditionally, brand associations are measured by analyzing the text data from consumers' responses to the survey or their online conversation logs. In this paper, we go beyond textual media and take advantage of large-scale photos shared on the Web. More specifically, we jointly achieve the following two fundamental tasks in a mutuallyrewarding way: (i) detecting exemplar images as key visual concepts associated with brands, and (ii) localizing the regions of brand in images. For experiments we collect about five millions of images of 48 brands crawled from five popular online photo sharing sites. We then demonstrate that our approach can discover complementary views on the brand associations that are hardly obtained from text data. We also quantitatively show the superior performance of our algorithm for the two tasks over other candidate methods.