I'm trying to make OpenCV use GPU on google Colab but I can' find any tutorial what I fond is this tutorial
but I get stuck at step3 cuz this tutorial is not for colab
-
Check this out: medium.com/deep-learning-turkey/…Axiumin_– Axiumin_2020-08-10 17:25:39 +00:00Commented Aug 10, 2020 at 17:25
-
I installed OpenCV like the tutorial but it didn't use the GPUall_alone– all_alone2020-08-10 17:36:38 +00:00Commented Aug 10, 2020 at 17:36
-
@Axiumin_ that tutorial, great as it is, is for using the GPU with colab, not what the question requiresKansaiRobot– KansaiRobot2021-05-27 03:00:47 +00:00Commented May 27, 2021 at 3:00
-
@all_alone I haven't tried it but it seems that in that tutorial, the standard version of opencv is installed, therefore CUDA is not installed. The tutorial does not provide a solutionKansaiRobot– KansaiRobot2021-05-27 03:01:48 +00:00Commented May 27, 2021 at 3:01
Add a comment
|
1 Answer
the default cv2 version installed on colab is 4.1.0 (very old !) and indeed, it has no CUDA support builtin. however, this is a google machine, it has all the tools, nooks & crannies to build your own ! latest cv2 ! with cuda ! (might take a good two hrs, though)
## 1st: switch to a gpu enabled runtime (from menu)
!git clone https://github.com/opencv/opencv
!git clone https://github.com/opencv/opencv_contrib
!mkdir /content/build
%cd /content/build
!cmake -DOPENCV_EXTRA_MODULES_PATH=/content/opencv_contrib/modules \
-DBUILD_SHARED_LIBS=OFF \
-DBUILD_TESTS=OFF \
-DBUILD_PERF_TESTS=OFF \
-DBUILD_EXAMPLES=OFF \
-DWITH_OPENEXR=OFF \
-DWITH_CUDA=ON \
-DWITH_CUBLAS=ON \
-DWITH_CUDNN=ON \
-DOPENCV_DNN_CUDA=ON \
/content/opencv
!make -j8
## the new cv2.so is now in /content/build/lib/python3/
## you are not allowed a normal install on colab, (not root)
## but you can copy it into your work folder and restart the runtime, so it forgets about the old version !
import cv2
cv2.__version__
!ls -l
## save for later use:
!cp /content/build/lib/python3/cv2.cpython-37m-x86_64-linux-gnu.so "/content/drive/My Drive/cv2_cuda"
## next time, load it into your work folder:
## dont forget to restart the runtime, so it forgets about the old version !
!cp "/content/drive/My Drive/cv2_cuda/cv2.cpython-37m-x86_64-linux-gnu.so" .
