README

For this project we developed multiple models:
- Model emsemble -> to test SciKit Learn simple models on dataset
  - python model_ensemble.py

- Simple model -> a 3 layer simple neural network
  - python simple_model.py

- Deep model -> convolutional neural network 
  - python deep_model.py

You can use the deep model to predict the ASL letter you are signing by running the cv-deep-model.py file: python cv-deep-model.py

Saved models can be used in this CV funcionality by specifying the model of choice.
- asl_model.pth -> just the trained deep model
- asl_model_lr_flip.pth -> flips half of initial data so this model is better at detecting both hands

The opencv-hand-tracking folder simply demos the MediaPipe hand tracking.
The live-demo folder will allow you to create your own dataset of images and train a much more accurate model for yourself.
This time you will use a random forest classification model trained on your images to predict ASL letters via webcam.
This demo will use one of the saved models from the models folder.