The repo is organized in sections. The main code right now runs a training on the bodies at rest data with three labelled classes supine, lateral, and prone.
Project constants file with all the paths and directories
Changing the dataset to have the 8 original labels found in bodies at rest
Preprocessing to account for imbalance data
Split the training randomly into validation data set by physically moving the files
Make the training more modular by allowing # of layers, type of layers to be defined
Classes for different datasets that will be used