This is an implementation to filter out the wrong face images in a certain directory filled with face images from same person, most of the time generated by StyleGAN. To be more specific, there are some exceptional cases when images generated by StyleGAN has poor quality, indicating that it lacks property to be identified as a 'human face'. Also, this implementation aims to filter out the face image of a different person as well.
Face Recognition is a very useful opensoure for face detection built on deep learning. Along with face detection, they also supports face verificaion function as well. Here, I used their face_distance function that first encodes the images and then compare the distance between these encoded vectors.