Social re-ranking of image based on visual and semantic information

Abstract

Social media sites like Flickr permit clients to annotate pictures with free labels referred as "tag", which altogether contribute to the improvement of the web picture retrieval. Tag-based picture hunt is an important strategy to discover pictures contributed by social media users in such social sites. In this paper, we propose a social re-positioning framework for tag-based and feature based image retrieval with the thought of image significance and assorted qualities. In social media network, measuring the relevance of tags with respect to the visual, semantic and view contents is a critical task. We go for re-positioning image, concurring to their visual data, semantic data and social pieces of information. This framework gives the potential of tag relevance fusion for real-world deployment.