Project to find the regions of interest in the face and mask them from a stream of images / video.
The project is currently setup to find the ROIs forehead, left cheek and right cheek, to change the ROIs use the reference of the Canonical face model image which is taken from the mediapipe repo.
Depends on numpy, opencv, imageio, h5py and mediapipe.
pip install numpy opencv-python mediapipe imageio h5py
Use either the get_face_roi function or draw_roi functions from face_roi.py.
from face_roi import display_roi, get_face_roi, mat_loader, mat_converter
Also note that using mat_converter and display_roi functions in the same file is not working as expected. It is recommended to use something like converter.py to convert the mat files beforehand.
main.py works, but it is just a proof of concept with little functionality.
Examples are provided in the examples notebook.