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 (%run -i main.py in ipython console)

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:

1. download the pretrained models here:
   • cls.pth (required)
   • cae.pth (optional)
   • drag and drop the models in resources/models/

OR

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