motor-imagery-bci-1-acquisition.xml

Working with OpenViBE signal processing scenarios and doing scenario/BCI design
Post Reply
mhadji05
Posts: 53
Joined: Tue Mar 16, 2021 9:37 am

motor-imagery-bci-1-acquisition.xml

Post by mhadji05 »

Hello all,

I' m trying to run the motor-imagery-bci-1-acquisition.xml scenario and I want make the imagination of left/right hand movement without of eeg device but with prerecorded data. Any idea how can I do this? Is there any suitable data-set for this?

Thanks.

Thibaut
Posts: 264
Joined: Wed Oct 31, 2018 9:14 am

Re: motor-imagery-bci-1-acquisition.xml

Post by Thibaut »

HI,
The acquisition scenario is used for..... acquisition^^, you train your classifier with the next scenario : train. An example eeg file is already set in this scenario
Thibaut

mhadji05
Posts: 53
Joined: Tue Mar 16, 2021 9:37 am

Re: motor-imagery-bci-1-acquisition.xml

Post by mhadji05 »

I have trained the classifier successfully. The .cfg file has been generated.

Also when the classifier is trained, I want to replay some recorded test data and depending of the result, print a left or right message in Unity 3d.
How can I achieve this? Is there some instructions for something like that or can someone give me these instructions?

I would appreciate any help!

Thibaut
Posts: 264
Joined: Wed Oct 31, 2018 9:14 am

Re: motor-imagery-bci-1-acquisition.xml

Post by Thibaut »

HI,
I have down some hello world game with Unity
See : viewtopic.php?p=16495#p16495
viewtopic.php?p=16511#p16511
Thibaut

mhadji05
Posts: 53
Joined: Tue Mar 16, 2021 9:37 am

Re: motor-imagery-bci-1-acquisition.xml

Post by mhadji05 »

Hello,

When i run this scenario at the end i take Accuracy = 0.0?

Why is this happening and why generated confusion matrix in the acquisition phase?

RESULTS:
[ INF ] At time 430.594 sec <Box algorithm:: (0x02e67945, Ox5ea8d309) aka Classifier trainer> Finished with partition 6 / 7 (performance : 26.5372173)
[ INF ] At time 430.594 sec <Box algorithm:: (0x02e67945, Ox5ea8d309) aka Classifier trainer> Finished with partition 7 / 7 (performance : 31.0679613)
[ INF ] At time 430.594 sec <Box algorithm: : 0x02e67945, Ox5ea8d309) aka Classifier trainer> Cross-validation test accuracy is 31.573362% (sigma = 6.9795798)
[ INF ] At time 430.594 sec <Box algorithm:: (0x02e67945, Ox5ea8d309) aka Classifier trainer> Cls vs cls 1 2
[ INF ] At time 430.594 sec <Box algorithm: : 0x02e67945, Ox5ea8d309) aka Classifier trainer> Target 1: 42.1 57.9, 1080 examples
[ INF ] At time 430.594 sec <Box algorithm:: (0x02e67945, Ox5ea8d309) aka Classifier trainer> Target 2: 79.0 21.0 %, 1080 examples
[ INF ] At time 430.594 sec <Box algorithm:: (0x02e67945, Ox5ea8d309) aka Classifier trainer Training set accuracy is 48.379630$ (optimistic)
[ INF ] At time 430.594 sec <Box algorithm:: (0x02e67945, Ox5ea8d309) aka Classifier trainer> Cls vs cls 1 2
[ INF ] At time 430.594 sec <Box algorithm:: (0x02e67945, Ox5ea8d309) aka Classifier trainer> Target 1: 51.4 48.6 %, 1080 examples
| INF | At time 430.594 sec <Box algorithm:: (0x02e67945, Ox5ea8d309) aka Classifier trainer> Target 2: 54.6 45.4 $, 1080 examples
! INF ] At time 0.000 sec <Box algorithm:: (0x00000de5, 0x0000613d) aka Graz Motor Imagery BCI Stimulator> Lua script terminated
[ INF ] At time 431.273 sec <Box algorithm:: (0x42570b5c, OxObba079c) aka Graz visualization> Confusion Matrix :
000 000
000 000
Accuracy = 0.0

Thibaut
Posts: 264
Joined: Wed Oct 31, 2018 9:14 am

Re: motor-imagery-bci-1-acquisition.xml

Post by Thibaut »

Hi,
It's normal, Graz visualization Confusion matrix works with classifier processor not trainer. Launch all scenario of the folder to see the protocol.
Thibaut

mhadji05
Posts: 53
Joined: Tue Mar 16, 2021 9:37 am

Re: motor-imagery-bci-1-acquisition.xml

Post by mhadji05 »

How can I add a 3rd class of Resting state? I want to train my classifier recognizing left-right hand movement and resting state if the user don't trying to imagine anything.

Thibaut
Posts: 264
Joined: Wed Oct 31, 2018 9:14 am

Re: motor-imagery-bci-1-acquisition.xml

Post by Thibaut »

Hi,
Right Click on box classifier trainer, inputs->new. I have never test the behavior on 3 class for this box, so I can help you so much.^^
Thibaut

Post Reply