motor-imagery-deep-learning

Implementation of Convolutional Recurrent Neural Network (CRNN) to decode motor imagery EEG data.

The implementation is based on the article: Cascade and Parallel Convolutional Recurrent Neural Networks on EEG-based Intention Recognition for Brain Computer Interface Dalin Zhang, Lina Yao, Xiang Zhang, Sen Wang, Weitong Chen, Robert Boots, Boualem Benatallah

Getting started

• Make sure you have Conda installed

• Create a isolated environment:

  $ conda create --name motor-imagery-deep-learning python=3.7
  

• Activate the environment:

  $ conda activate motor-imagery-deep-learning
  

• Install the dependencies:

  pip install -U -r requirements.txt
  

Modules

/data/dataset/{dataset-name}
    description.pdf: Optional pdf with a description of the dataset
    montage.png: Optional image of the montage used to the acquisition of the EEG data
    README.md: README about the dataset and the pre-processing steps

/datasets
    {dataset-name}.py: Script to load dataset

/models/{dataset-name}
    Package of scripts with models to resolve the dataset problem.

    README.md: README with description about each model in this package

/preprocessing/{dataset-name}
    Package of scripts to perform the pre-processing steps referring to dataset