Censoring Sensitive Data from Images

In the recent years, the vast volume of digitalimages available enabled a large range of learning methods tobe applicable, while making human input obsolete for manytasks. In this paper, we are addressing the problem of removingprivate information from images. When confronted with arelatively big number of pictures to be made public, one mayfind the task of manual editing out sensitive regions to beunfeasible. Ideally, we would like to use a machine learningapproach to automate this task. We implement and comparedifferent architectures based on convolutional neural networks, with generative and discriminative models competing in anadversarial fashion.