P3 system using only BCI2000 classifier
Posted: 31 Aug 2015, 18:13
Hello there,
I have had a lot of trouble using 64bit windows system, g.HIamp and BCI2000 for acquisition so am looking for alternative solutions. I have a user interface/application written in python that I would like to use. I have a custom python data acquisition system that can stream data and run python code in real time. I'm at the stage of writing my own LDA decoder in python but would rather not reinvent the wheel. I am using python 2.7 primarily and have seen that BCPy2000 doesn't work with this and don't think this the best solution.
I am looking for advice in using BCI2000/matlab code in my system to do the classification. I am pretty confident I can manipulate the streaming data but would like any advice in how best to use existing BCI2000 tools to do the training, format the stepwisefit() output for use in real-time classification and send the result back to a python program. I saw some suggestions on using 'MATLAB signal processing' instead of UDP between python and MATLAB. My C knowledge is very rusty but would be open to wrapping BCI2000 C++ but haven't done this before. Open to suggestions. I'm sure there is useful info I've left out so shoot questions.
I have had a lot of trouble using 64bit windows system, g.HIamp and BCI2000 for acquisition so am looking for alternative solutions. I have a user interface/application written in python that I would like to use. I have a custom python data acquisition system that can stream data and run python code in real time. I'm at the stage of writing my own LDA decoder in python but would rather not reinvent the wheel. I am using python 2.7 primarily and have seen that BCPy2000 doesn't work with this and don't think this the best solution.
I am looking for advice in using BCI2000/matlab code in my system to do the classification. I am pretty confident I can manipulate the streaming data but would like any advice in how best to use existing BCI2000 tools to do the training, format the stepwisefit() output for use in real-time classification and send the result back to a python program. I saw some suggestions on using 'MATLAB signal processing' instead of UDP between python and MATLAB. My C knowledge is very rusty but would be open to wrapping BCI2000 C++ but haven't done this before. Open to suggestions. I'm sure there is useful info I've left out so shoot questions.