Implementation of facial landmarks detection using pytorch on iBUG 300W dataset. Why Xception Net? Because it provides satisfactory accuracy, with least computational requirement and memory utilization, I mean can you believe that my model, which looks kinda massive, is only of 10MB!
Yet I am working on project with dlib 68 face landmarks. But its not as stable as this module. I am using 68_face_landmarks.dat file in my project. But can you guide how can I make ".dat" file with this module. Or can you please share ".dat" file for this module.