Start a FaceTime session when you blow a kiss!
First, capture the training data. Use normal or kiss to denote the label. I captured ~1,000 photos (500 for kiss and normal each)
and achieved 90% prediction correctness. So you may also need that more.
$ python camera_reader.py <label>
Second, train the model.
$ python model_train.py
Finally, run the script and try holding your gesture for a few seconds.
$ python camera_reader.py
Install OpenCV, Anaconda.
Build an AppleScript with following code and put it somewhere appropriate (like your desktop)
set phone_num to "THE_PHONE_NUMBER_TO_CALL"
set faceTime to "FaceTime"
if application faceTime is running then
return
end if
do shell script "open facetime://" & quoted form of phone_num
tell application "System Events"
repeat while not (button "Call" of window 1 of application process faceTime exists)
delay 1
end repeat
click button "Call" of window 1 of application process "FaceTime"
end tell