Full video: AR Masks
First the program detects user's face.
Once the face is detected, It uses that as an input for the trained Deep-learning model.
That model determines the expression on the face.
Once the expression is detected, the program then applies selected MASK with detected expreddion on the face.
As the expression changes, the mask expression also changes.
Using my hands, I can switch between diffrent mask designs.
The current mask design is shown in the top of the video frames.
Using SPACEBAR I can take "screenshots" of the image, which will be stored in "Saved Image" folder.
expression_recognition.py - Main file.
Masks folder - This contains all the images of masks. This folder must be present along with expression_recognition.py file.
Models folder - This contains the trained deep-learning model. The required model for this project is 4Emotions.h5 and there are other model which is trained over 7 diffrent emotions.
haarcascade_palm.xml - This haar-cascade file is used to detect palms in the frame.
shape_predictor_81_face_landmarks.dat - This file is used to project the 81 landmark points on the face.