/Muse-MotorImageryClassification

Record EEG data from a Muse 2 headband using the MInd Monitor app and python osc module. Build and train a CNN model in Keras framework to classify Left-Right Motor Imagery. Make real-time predictions using the trained model.

Primary LanguageJupyter NotebookMIT LicenseMIT

Muse-MotorImageryClassification

The project is aimed at creating a Real-time Left-Right Motor Imagery Classifier using CNN. Device used is a Muse 2 brain sensing headband with 4 electrode channels - TP9, AF7, AF8, TP10. Data is streamed using the 'Mind Monitor' app.

Following codes have been implemented till now:

MotorImagery_OSC_Record -

  1. Record and save EEG data as CSV files from a Muse 2 headband using the MInd Monitor app and python osc module.
  2. Events can be configured in the rec_dictionary

MotorImagery_Training -

  1. Configure and train a CNN model based on 'EEG-ITNet'
  2. Load the CSV data recordings into a Pandas dataframe and convert into MNE epochs for training

MotorImagery_OSC_Predict -

  1. Make real time predictions using the trained model.