Content-based Graph Privacy Advisor
This work addresses privacy concerns for users uploading images online, though it is incremental as it builds on an existing graph model.
The paper tackles the problem of predicting image privacy risks by developing Graph Privacy Advisor, which uses scene information and object cardinality to classify images as private, achieving improved performance through refined feature extraction and reduced model complexity.
People may be unaware of the privacy risks of uploading an image online. In this paper, we present Graph Privacy Advisor, an image privacy classifier that uses scene information and object cardinality as cues to predict whether an image is private. Graph Privacy Advisor simplifies a state-of-the-art graph model and improves its performance by refining the relevance of the information extracted from the image. We determine the most informative visual features to be used for the privacy classification task and reduce the complexity of the model by replacing high-dimensional image feature vectors with lower-dimensional, more effective features. We also address the problem of biased prior information by modelling object co-occurrences instead of the frequency of object occurrences in each class.