Review of methods to predict social image interestingness and memorability

Abstract

An entire industry has developed around keyword optimization for ad buyers. However, social media landscape has shift to a photo driven behavior and there is a need to overcome the challenge to analyze all this large amount of visual data that users post in internet. We will address this analysis by providing a review on how to measure image and video interestingness and memorability from content that is tacked spontaneously in social networks. We will investigate current state-of-the-art of methods analyzing social media images and provide further research directions that could be beneficial for both, users and companies.