Automated Face Detection And Swapping In Video

Face was used as one of the normal behaviors for identifying humans. It generally stays a challenge to cause a computer framework to carry on like how a human senses things. We have made our application to help and simulate the visualization of a human’s imagination of Deep Fake videos. The entire face swapping project is divided into three stages, the first stage extracts the frames from a video file. The Second stage includes face detection and landmark extraction from frames for which we have used the getfrontalface() and predictor() function of dlib library in Python which tells us whether a face is present in an image and also provides us the landmarks of the facial features. In the third stage, we use Delaunay triangulation and affine transformation to find the triangles by their index values and the coordinates of the triangles then these triangles are matched and swapped from one face to another. Finally, the application will generate frames with swapped faces, which are further combined into a video file.