This model blurs the faces of people in an image to preserve privacy. It was trained on the “Labeled Faces in the Wild” dataset and tested on small and medium sized images. This is a great tool to mitigate privacy concerns when showing images which contain people in a public setting. Please note: though this model currently is optimized for caucasian men based solely on the training data set coverage, it will also work for other ethnicities and genders. A deep dive on the training data set to better understand the limitations is recommended.