This repository contains the basic handwriting decoder model, as well as the required utilities to create training and testing data. This prototype depends on the publicly available Stanford dataset, and would require the folder handwritingBCIData to be copied into this root directory. The packages installed, with specific versions where necessary, are contained in the environment.yml file and can be used with conda to recreate the environment used in testing.
To use the files, first run the createTrainingData.py script in the src subdirectory, and then run the modelTraining.py script in the example subdirectory. Similarly, for the inference example, run the createTestingData.py scipt in the src subdirectory, and then run the modelInference.py scipt in the example subdirectory.