Image Privacy Prediction Using Deep Features

October 19, 2016
2:00 pm - 5:30 pm
Hall C

Track: General
Type: Posters
Level: All

Online image sharing on social media sites such as Facebook, Flickr can lead to unwanted disclosure and privacy violations, when privacy settings are used inappropriately. In this study, we explore deep visual features and tags for image privacy prediction. The results show that deep visual features outperform SIFT and GIST, also deep tags combined with user tags perform best among all features.

Speaker(s)

, Research Assistant, University of North Texas