Facial Feature Detection

Overview

`notebooks/` - showcases the process of the ML pipeline

1. extract_faces.ipynb - data cleaning
2. pytorch.ipynb - training, validation
3. eval.ipynb - testing, visualization of results

`main.py` - entrypoint for running the live facial feature detection

`webpage.py` - entrypoint for running the flask hosted webpage

Instructions

To run main.py or webpage.py, you need the model files. You can either:

download the pretrained models here:
- cae.pth (optional)
- cls.pth (required)

OR

download the CelebA dataset and train your own models by running the notebooks in order