due to some boredom, i threw together a small exaple how to use opencv and panda together to create a small application based on face-detection/head-tracking. it’s a bit jerky but works.
http://home.arcor.de/positiveelectron/files/Ball-in-Maze-headtracking.zip
you need to have the opencv-python installed to run it.
some people have issues even though they have it installed, looks like some strange version-problem in opencv itself.
i have not tested the code that much, it should run thought.
technical info: opencv and panda are running in seperate processes communicating via sockets. so if you have dualcore you’r lucky cause one core does the opencv tracking while the other one runs panda.
since there is no smoothing of the tracking data a well-lit room and a webcam with good fps is recommended or else it will be really jerky.
from Tut-Ball-in-Maze.py and launch facedetect.py from a console cos I’m on kubuntu I guess but that’s not a problem
EDIT
I forgot to say - for the records - that I installed the latest opencv release from sourceforge and plus I had issues installing it 'cos since I’m on AMD mobo and the default installer configuration is settled to use intel SSE2 therefore I’d to launch configure like this:
heh I had it before but Thomas’ script gave me errors so I take for grated he was using a new opencv release and so I installed it. Anyway now it’s all ok.
i heared about it. it was used with panda already. but its non-free so i had no further intrest.
a, from my point of view, more intresting projects it this one http://code.google.com/p/ehci/ since it offeres full 6DOF headtracking. also comes with a panda-sample
yeah, this project looks pretty good so far!!!
after i fixed my maya problems, i hope i will find time to play a bit around with the tracker.
btw. you can use a none commercial licence of the faceAPI.
jeah non-comercial ones… which limits you to:6DOF head tracking.
which is pretty much the same as the google-project too. no eyebrows, no mouth no nothing comes with the non-commercial license. + it prevents you to write commercial applications.
often its useful to see how otherones are doing there things and it could bring you on different ideas.
its like the game with the wheel and a wing, in the meantime we nearby are touching other planets with hands. the thoughts are getting bigger…
so im happy when i can take a look into a otherones cookbook, but sure i can understand you, if you know there are so many othergreat points in. it makes you sad that you cant see everything.but thats the trading game, noidea where this will lead us hope not into my bad imaginations.
Sorry to resurrect this post, I am trying to get this working… and I cant seem to find the right python-opencv package. I am using panda 1.6.2, windows XP.
What version of opencv is needed for this? Has anyone tried opencv v.2 ?
Panda is python 2.5 right?
If yes, can anyone provide a link for me to d/l it?
DirectStart: Starting the game.
Known pipe types:
wglGraphicsPipe
(all display modules loaded.)
:display:windisplay(warning): SetForegroundWindow() failed!
Traceback (most recent call last):
File “Tut-Ball-in-Maze.py”, line 27, in
from opencv import *
ImportError: No module named opencv
**** End of process output ****
i saved opencv.1.1pre1 to program files, do i have to save it to the panda3d folder and rename it to opencv? I tried it and it wouldn’t work
any ideas?
dunno if you need to copy anything around. in my case the linux package management took care of that. may help on windows,tho.
this is more a opencv-install issue i guess. i dont really know enough to help you with that.