Discovery of Elsagate: Detection of Sparse Inappropriate Content from Kids Videos

Abstract

Elsagate refers to kids videos containing inappropriate content for children but difficult to filter thus often shown up on kids channels. Most of the mainstream kids channels have powerful filters. However, when the inappropriate content is sparse, the filters often fail to detect the inappropriateness. In this paper, we introduce our work in progress, a scheme to detect Elsagate videos based on Sparse Linear Discrimination (SLD), an effective way to help detect and classify these kinds of unsafe videos and enrich better user experience.