Signal_Glove_Model

Official ML repo. for the sign language interpreter.

How to use

Data preprocessing

1. Put raw data in a folder called "raw_data", or edit your desired folder path in "config.json"
2. Run python format_dataset.py to re-format your data into a folder called "motion_data"
3. Run python generate_split.py to generate train/validation/testing split

Running model

1. To train a model, run python train_model.py with the evaluate_saved flag set to None.
2. To evaluate a trained model, run python train_model.py with the evaluate_saved flag appointed to your desired model path.