Besided python 3, these modules are required and can be installed using the
command pip3 install -r requirements.txt when inside the project root folder.
- scipy
- matplotlib
- sklearn
- numpy
- An edited python-ardrone library is needed. This can be cloned
or downloaded from https://github.com/DevinHillenius/python-ardrone and
manually installed using
python3 setup.py.
The eeg.py script allows users to visualise the powers of EEG signals using
scatterplots.
The folders containing the signals can be selected using the -f or --folder
option (see Examples). The sample rate can be adjusted using the -s or
--sample_rate flag, it is 256 by default (see Examples). The frequency band
is 8Hz to 13Hz by default and can be adjusted as well, using the -b or
--band flag. Again, see the Example section for its usage.
We organised our EEG data in the following way. The root folder data is divided
in multiple folders, each containing all the data about a particular
individual. The folders of the individuals are divided into multiple folders
each containing a single motor imagery movement, such as imagining moving
hand or feet. Inside these motor imagery movement folders are two files,
c1.mat and c2.mat. These are the channels of the actions and contain all
records of that particular action. c1.mat and c2.mat are required to be
present in the selected folders when using the eeg.py script. This data can be
found on a private repository and is not available to the public.
data\
personA
personB
personC\
foot-right-cond
hand-left-base
hand-left-cond
hand-right-cond\
c1.mat
c2.mat# Sample rate is 256 and band is 8 to 13 by default
./eeg.py --folder data/hand-left-cond data/foot-right-cond/
# Using 3 different actions, a different sample rate and a different band
./eeg.py --folder data/hand-left-cond data/hand-right-cond data/foot-right-cond/ --sample_rate 512 --band 3 8