Places with tutorials to make my RPi run OpenCV and start pattern recognition include the part where I have to be Building OpenCV.



I mean OpenCV is already in these pips that can be placed in virtualenvrionments from pip install and be developed from env directory. no?

Or is it because my RPi has to adapt to OpenCV so it needs to figure it out? -And that is why figuring it out or building takes like 4 hours? But can it be that these guys have thought of availability for others to work on my version of OpenCV because it is the one to be uploaded on github in case of betterment changes?

My priority goes to integrating this as an Internet Thing. Have it its own API as it recognizes. Would SimpleCV be less resourceful for prototyping?

1 Answer 1


OpenCV is like many open-source projects a moving target. To get the latest bug-fixes and features , you have to build (compile) it from source , especially since the versions shipped with Debian/Raspbian tend to be ... older.

Python modules can be written as wrappers around compiled code. OpenCV provides exactly such a Python module , it is based around OpenCVs C++ core.

The official OpenCV wrappers/bindings for Python don't seem to be available via pip , but get installed as part of the greater cmake-based compilation/installation process itself.

  • Let me wind it up. So because pip does not include the newest version of openCV, I should use the version that has the newest stuff on it therefor I then have to get it from git and compile it with the said cmake. I can also use pip, but then I am not getting the newest version even if that takes less time to be set up for coding.
    – bgut
    Mar 30, 2017 at 11:59
  • 1
    I rephrased my answer.
    – flakeshake
    Apr 2, 2017 at 12:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.