Using OpenCV recognize-faces.py allows users to tell the computer their name and the computer will then train a model to detect their face live on the webcam. Training data is caputured using OpenCV's Haar Cascade model and the final model is trained on Local Binary Patterns Histograms.
Add an algorithm to rotate the faces for the live feed predictions. Although this would slow computation it would allow for much more accuracy because the model currently has trouble recognizing faces that are sideways.