A Systematical Solution for Face De-identification

With the identity information in face data more closely related to personal credit and property security, people pay increasing attention protection of privacy. In different tasks, have various requirements for de-identification (De-ID), so we propose a systematical solution compatible these De-ID operations. Firstly, an attribute disentanglement generative network is constructed encode two parts face, which are (facial features like mouth, nose eyes) expression (including expression, pose illumination). Through swapping, can remove original ID completely. Secondly, add adversarial vector mapping perturb latent code image, from previous traditional methods. this, construct unrestricted image decrease similarity recognized by model. Our method flexibly de-identify ways processed images high quality.