Using EEG collected from OpenBCI boards for emotion recognition model

amirajamiraj UCI
in Ganglion

Hi everyone,

We are a research group at UCI. We plan to collect EEG to have a EEG based emotion recognition model.
We are looking for a cost efficient tool so we can easily collect EEG signals from it.
Also we want to make sure that EEG collected is good enough to have a good accuracy for our model.
We have muse 2 headband currently, but it seems we cannot quite trust EEG collected from it although working with it is really easy.
Is OpenBCI boards like Ganglion good option for our research? Do you have any recommendation on this?

Thanks for your time and considerations.
Best,
Amir

Comments

  • wjcroftwjcroft Mount Shasta, CA

    Amir, hi.

    See this paper by Jeremy Frey comparing Cyton with a medical grade EEG from g.tec:

    https://arxiv.org/abs/1606.02438

    As far as other research, papers employing OpenBCI, see the citation list:

    https://docs.openbci.com/citations
    https://docs.openbci.com/docs/Welcome.html

    Regards, William

  • retiututretiutut Louisiana, USA

    @amiraj You can even use the Ganglion+WiFi shield to stream data at 1600hz! This is the recommended configuration for the WiFi shield.

  • BillhBillh

    You can get sketchy and not consistent between subjects emotional valence data from parietal electrodes (P4 and P3), but I believe the facial expression analysis literature shows better sensitivity and specificity than the EEG data.

    So if you are trying to use machine learning to judge reactions to events on a screen I'd use take facial video data over EEG data.

  • amirajamiraj UCI

    @wjcroft said:
    Amir, hi.

    See this paper by Jeremy Frey comparing Cyton with a medical grade EEG from g.tec:

    https://arxiv.org/abs/1606.02438

    As far as other research, papers employing OpenBCI, see the citation list:

    https://docs.openbci.com/citations
    https://docs.openbci.com/docs/Welcome.html

    Regards, William

    Thank you William... I was looking for a paper like this. Appreciate it.

  • amirajamiraj UCI

    @retiutut said:
    @amiraj You can even use the Ganglion+WiFi shield to stream data at 1600hz! This is the recommended configuration for the WiFi shield.

    Thank you to mention. Speed of streaming data is important for us. 1600Hz is a lot... compared to 200Hz we had before.

  • amirajamiraj UCI

    @Billh said:
    You can get sketchy and not consistent between subjects emotional valence data from parietal electrodes (P4 and P3), but I believe the facial expression analysis literature shows better sensitivity and specificity than the EEG data.

    So if you are trying to use machine learning to judge reactions to events on a screen I'd use take facial video data over EEG data.

    Thank you billh.
    We also have plan to also add facial expression in our study too.

Sign In or Register to comment.

© Vanilla Keystone Theme 2026