Simple guide for installing opencv 4.7.0 on Ubuntu with Cuda and cuDNN.
(To install opencv with cuda on NVIDIA Jetson Nano, skip to the end of this guide)
There are 4 main components of installing opencv with Cuda on Ubuntu.
- Install-Opencv-4.7-CUDA-Ubuntu
- To install opencv with cuda on jetson nano:
Open the CUDA download link.
In this example, the related NVIDIA driver version is 535 which is present in the file named:
"cuda-repo-ubuntu2004-12-2-local_12.2.0-535.54.03-1_amd64.deb"
We will need to install this driver version which will be shown in the next steps.
After performing the operations above, you will have the CUDA installed.
The official guide: guide link
You directly can use this link to reach the following bash commands:
To find cuDNN_version and cuda_version, use tab button to autocomplete.
Example bash commands for cuDNN 8.9.2 (CUDA 12.X):
sudo apt-get install libcudnn8=8.9.2.26-1+cuda12.1 \
sudo apt-get install libcudnn8-dev=8.9.2.26-1+cuda12.1 \
sudo apt-get install libcudnn8-samples=8.9.2.26-1+cuda12.1
After the installation, you can check if the cuDNN works properly by following the steps in this link.
!!(Do not go through this steps if you did the network installation)!!
To download cuDNN in tar version and install, it is necessary to have a NVIDIA account. After you sign in, go to the download link and download the related cuDNN version:
When the download is finished, follow the installation steps:
After the installation, you can check if the cuDNN works properly by following the steps in this link.
Before that, you should download the cuDNN samples with the following bash command in this part starting with "libcudnn8-samples".
For completing the cuDNN and cuda versions in shell, use tab button to autocomplete in desired versions.
Search for the software & update application which is shown in the image below (the purple colored one).
Then, go to the "Additional Drivers" section.
Select the correct driver version (also do not install server drivers) and apply changes.
After the driver installation, reboot the system. Then run the following command:
nvidia-smi
If you see the correct version installed in the screen, you are done with the GPU driver installation.
Note: The CUDA version in the top-right corner of the screen is not showing the version your system has. It shows the most recent supported CUDA version for your driver. If you want to see the CUDA version, run the nvcc --version command.
For this part, you have to install cmake and a proper C++ compiler.
For this purpose, run the following commands:
$ sudo apt install cmake
$ sudo apt install gcc g++
If you haven't installed python yet, then install it and the numpy library:
$ sudo apt install python3 python3-dev python3-numpy
Install the other dependencies:
\
sudo apt-get install build-essential cmake pkg-config unzip yasm git checkinstall
sudo apt-get install libjpeg-dev libpng-dev libtiff-dev libopenjp2-7-dev
sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev
sudo apt-get install libgstreamer1.0-dev libgstreamer-plugins-base1.0-dev
sudo apt-get install libxvidcore-dev x264 libx264-dev libfaac-dev libmp3lame-dev libtheora-dev
sudo apt-get install libfaac-dev libvorbis-dev
sudo apt-get install libopencore-amrnb-dev libopencore-amrwb-dev
sudo apt-get install libgtk-3-dev
sudo apt-get install libtbb-dev
sudo apt-get install libatlas-base-dev gfortran
sudo apt-get install libprotobuf-dev protobuf-compiler
sudo apt-get install libgoogle-glog-dev libgflags-dev
sudo apt-get install libgphoto2-dev libeigen3-dev libhdf5-dev doxygen
sudo apt-get install libgtkglext1 libgtkglext1-dev
sudo apt-get install libopenblas-dev liblapacke-dev libva-dev libopenjp2-tools libopenjpip-dec-server libopenjpip-server libqt5opengl5-dev libtesseract-dev
sudo apt-get install build-essential cmake git unzip pkg-config zlib1g-dev
sudo apt-get install libjpeg-dev libjpeg8-dev libjpeg-turbo8-dev
sudo apt-get install libpng-dev libtiff-dev libglew-dev
sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev
sudo apt-get install libgtk2.0-dev libgtk-3-dev libcanberra-gtk*
sudo apt-get install python-dev python-numpy python-pip
sudo apt-get install python3-dev python3-numpy python3-pip
sudo apt-get install libxvidcore-dev libx264-dev libgtk-3-dev
sudo apt-get install libtbb2 libtbb-dev libdc1394-22-dev libxine2-dev
sudo apt-get install gstreamer1.0-tools libgstreamer-plugins-base1.0-dev
sudo apt-get install libgstreamer-plugins-good1.0-dev
sudo apt-get install libv4l-dev v4l-utils v4l2ucp qv4l2
sudo apt-get install libtesseract-dev libxine2-dev libpostproc-dev
sudo apt-get install libavresample-dev libvorbis-dev
sudo apt-get install libfaac-dev libmp3lame-dev libtheora-dev
sudo apt-get install libopencore-amrnb-dev libopencore-amrwb-dev
sudo apt-get install libopenblas-dev libatlas-base-dev libblas-dev
sudo apt-get install liblapack-dev liblapacke-dev libeigen3-dev gfortran
sudo apt-get install libhdf5-dev libprotobuf-dev protobuf-compiler
sudo apt-get install libgoogle-glog-dev libgflags-dev
Then, download the opencv and opencv contrib folders for the version 4.7.0
cd ~
wget -O opencv.zip https://github.com/opencv/opencv/archive/4.7.0.zip
wget -O opencv_contrib.zip https://github.com/opencv/opencv_contrib/archive/4.7.0.zip
unzip opencv.zip
unzip opencv_contrib.zip
mv opencv-4.7.0 opencv
mv opencv_contrib-4.7.0 opencv_contrib
rm opencv.zip
rm opencv_contrib.zip
Now, you should have opencv and opencv_contrib folder in your home directory.
So, we are ready to build opencv with CUDA. Make a build folder in opencv folder.
cd ~/opencv
mkdir build
cd build
After these commands, you are in the build folder. To run the cmake command, use the appropriate flags.
IMPORTANT: Change the XXXXX with your GPU's current CUDA architecture which you can find in this link in the Compute Capability name with the X.X format (such as 7.5 or 8.6).
Also, you can turn "-D WITH_QT=ON" to OFF if you do not want to build with QT.
cmake -D CMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D OPENCV_EXTRA_MODULES_PATH=~/opencv_contrib/modules \
-D EIGEN_INCLUDE_PATH=/usr/include/eigen3 \
-D WITH_OPENCL=OFF \
-D WITH_CUDA=ON \
-D CUDA_ARCH_BIN= XXXXX \
-D CUDA_ARCH_PTX="" \
-D WITH_CUDNN=ON \
-D WITH_CUBLAS=ON \
-D ENABLE_FAST_MATH=ON \
-D CUDA_FAST_MATH=ON \
-D OPENCV_DNN_CUDA=ON \
-D WITH_QT=ON \
-D WITH_OPENMP=ON \
-D BUILD_TIFF=ON \
-D WITH_FFMPEG=ON \
-D WITH_GSTREAMER=ON \
-D WITH_TBB=ON \
-D BUILD_TBB=ON \
-D BUILD_TESTS=OFF \
-D WITH_EIGEN=ON \
-D WITH_V4L=ON \
-D WITH_LIBV4L=ON \
-D WITH_PROTOBUF=ON \
-D OPENCV_ENABLE_NONFREE=ON \
-D INSTALL_C_EXAMPLES=OFF \
-D INSTALL_PYTHON_EXAMPLES=OFF \
-D PYTHON3_PACKAGES_PATH=/usr/lib/python3/dist-packages \
-D OPENCV_GENERATE_PKGCONFIG=ON \
-D BUILD_EXAMPLES=OFF ..
If the configuring is done without any problem, you are ready to run the make command.
Change the 8 with the number of cores in your CPU that you want to utilize in the build process.
make -j8
After the make in done, you are ready to install.
sudo make install
sudo ldconfig
sudo apt-get update
To check, you can open the python interpreter, and run:
import cv2
print(cv2.getBuildInformation())
print(f"The opencv version is: {cv2.__version__}")If you see the correct version intalled with the CUDA and cuDNN, you are done with the opencv installation!
If you are trying to build opencv 4.7 with cuda on jetson nano, you can look at this bash script and use it part by part or directly run it, or you can have a look at this tutorial.
WARNING: You have to increase your nano's swap memory in order to build using 4 cores.