error: (-216:No CUDA support)

[hadign20]

System information

• OpenCV => 4.5.3
• Operating System / Platform => Windows 64 Bit
• Compiler => Visual Studio 2019
• GPU ==> NVIDIA GeForce GTX 745 (compute capability: 5.0)
• CUDA version ==> 10.2
• CUDNN version ==> 8.2.0

Detailed description

I cannot use opencv with cuda. I have compiled and built opencv from source with cmake. Every time I try to use GPU with opencv I get this error message:

OpenCV(4.5.3) C:\opencv\opencv4.5.3\opencv\modules\dnn\src\dnn.cpp:2385: error: (-216:No CUDA support) OpenCV was not built to work with the selected device. Please check CUDA_ARCH_PTX or CUDA_ARCH_BIN in your build configuration. in function 'cv::dnn::dnn4_v20210608::Net::Impl::initCUDABackend'

This is the cmake configuration:

cmake ^ -G "Visual Studio 16 2019" ^ -A x64 ^ -DOPENCV_EXTRA_MODULES_PATH=%cwd%/opencv_contrib/modules ^ -DWITH_CUDA=ON ^ -DWITH_CUDNN=ON ^ -DOPENCV_DNN_CUDA=ON ^ -DWITH_CUBLAS=ON ^ -DWITH_CUBLAS=ON ^ -DCUDA_ARCH_BIN=7.5 ^ -DBUILD_opencv_world=ON ^ -DBUILD_opencv_python3=ON ^ -DPYTHON_EXECUTABLE=C:/ProgramData/Anaconda3/python.exe ^ -DPYTHON_INCLUDE_DIRS=C:/ProgramData/Anaconda3//include ^ -DPYTHON_LIBRARY=C:/ProgramData/Anaconda3//libs/python38.lib

and this is the output of CMAKE:

I also tried with -DCUDA_ARCH_BIN=5.0 because my GPU CC is 5.0, but got the same error. How can I fix this issue?

[xlabd]

I think the installation error should now be fixed. Ref: https://github.com/opencv/opencv/issues/20706

@vikasgupta-github we can close it

[hadign20]

Yes, the problem is OpenCV DNN doesn't work with NVidia GPU. I found that out the hard way.

If you have NVidia GPU try deep learning with other platforms, such as TensorFlow.

[xlabd]

If you check this blog, specifically the last section, you can see how the dnn module speeds up the inference with the help of gpu support.

[Hyrtsi]

I had the same problem. It was because I built it with the wrong CC indeed. However, I was also missing this:

CUDA_GENERATION=Auto

So my whole command is

cmake -D CMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D WITH_TBB=ON \
-D ENABLE_FAST_MATH=1 \
-D CUDA_FAST_MATH=1 \
-D CUDA_GENERATION=AUTO \
-D WITH_CUBLAS=1 \
-D WITH_CUDA=ON \
-D BUILD_opencv_cudacodec=OFF \
-D WITH_CUDNN=ON \
-D OPENCV_DNN_CUDA=ON \
-D CUDA_ARCH_BIN=8.6 \
-D CUDA_GENERATION=Auto \
-D WITH_V4L=ON \
-D WITH_QT=OFF \
-D WITH_OPENGL=ON \
-D WITH_GSTREAMER=ON \
-D OPENCV_GENERATE_PKGCONFIG=ON \
-D OPENCV_PC_FILE_NAME=opencv.pc \
-D OPENCV_ENABLE_NONFREE=ON \
-D OPENCV_EXTRA_MODULES_PATH=../opencv_contrib-4.x/modules ../opencv-4.x \
-D INSTALL_PYTHON_EXAMPLES=OFF \
-D INSTALL_C_EXAMPLES=OFF \
-D BUILD_EXAMPLES=OFF ..

I know the issue is marked to be closed but I figure this could help someone who had the same problem as I did :)

Also remember to rm -rf * in build directory before running the conf and build again just to make sure everything goes correctly.