This is an implementation of paper "Face Alignment at 3000 FPS via Regressing Local Binary Features" with Python.

300W dataset is used here for simplicity.

If you don't have the 300w_cropped dataset, run
    ./setup.sh
or if you already have it, you can do the steps after the downloading part of setup.sh

After dataset preparation, run
    python train.py
to do the training. Python 3 is used here.

After Training, run
    python test.py
to do the testing.