Geolocated Visual Summarization of Social Media Data

Abstract

The usage of social media is increasing day by day. People use social media platforms to communicate with their friends or other users and to demonstrate what they are interested in by sharing different kinds of media such as photos, texts, and videos. A portion of the posted content also include location information. Such posts having location information are called geo-tagged posts in social networks. According to the analysis of geo-tagged posts, popular locations or activities can be identified. This study proposes a method to identify the most representative subset of the visual content shared in a region through social media. Our approach aims to detect the popular places and events and utilizes Scale-Invariant Feature Transform (SIFT) features. Identified representative visuals are used to generate a web based tourist map. In this study, Flickr is used as the source of geotagged visual content.