Seeing is Believing: Analytics of Visual Data

Distribution

Presented at the 2013 Digital Analytics Association.

Abstract

Humans process extreme amounts of visual information every second, making visual artifacts (e.g., images) a natural, efficient, and effective form of communication.

Online social networks, coupled with the ubiquity of mobile devices and cameras, reflected this process: users communicate with images. Despite this, traditional analytics of social networks rely on text to describe what the users are communicating, largely due to the technological challenges surrounding the analysis of visual data.

In this talk, we will discuss how Curalate, a marketing platform for the visual web, is tackling the technological challenges around visual analytics using novel computer vision and big data techniques.

Specifically, we explore how even the foundational analytic techniques such as aggregation, prediction, and classification present significant challenges when applied to visual data, and how Curalate addresses these challenges.

Slides

[ pdf ]