System development of commercial logo analysis on online social media

Abstract

Social Network Analysis has becoming a new business recently. Many existing systems for social network analysis, however, are still limited to analyzing text. None of them analyzes the content of images that circulate on social media. In this paper we describe our effort in developing a system for analyzing the occurrence of commercial logos (company emblem). A photo-sharing social network is used as the source of images for analysis. An object detection and recognition algorithm is then applied to detect and recognize the occurrence of a logo in the retrieved images. The analysis of logo includes visualizing the trend of logo that has been posted over time as well as visualizing its spatial distribution over regions of interests. Our experiment on SIFT, SURF and PAST algorithms for detection and recognition of logo occurrence in image dataset reveals that the best performer is the Scale Invariant Feature Transform (SIFT) algorithm. We also perform usability testing on the developed system. The results show that our system is effective, easy to learn & use as well as being helpful to users.