opencv-cuda-ubuntu2204

Compile OpenCV with NVIDIA GPU CUDA support under Ubuntu 22.04 in a virtual environment

Note: These instructions can be used for a minimal Ubuntu installation option using Python 3.10.

Step 1: Remove any NVIDIA drivers

• sudo dpkg -P $(dpkg -l | grep nvidia | awk '{print $2}')

• sudo apt autoremove

Step 2: Install NVIDIA CUDA Drivers and Toolkit according to NVIDIA Data Center Documentation Software Deployment Workflow

Note: This installation has been tested using the 550 version drivers.

Install NVIDIA Kernel Open Source and CUDA Drivers:

• sudo apt-get install --verbose-versions nvidia-kernel-source-550-open cuda-drivers-550

Install the latest CUDA Toolkit using:

• sudo apt install cuda-toolkit

Enable persistence mode for the GPU to reduce power draw at idle:

• sudo nvidia-smi --persistence-mode=ENABLED

Step 2b (Optional): Install NVIDIA GPU DirectStorage (GDS) Drivers according to NVIDIA GPUDirect Storage Installation and Troubleshooting Guide

Note: Refer to website for pre-installation instructions, including IOMMU adjustments.

• sudo apt install nvidia-gds

Step 3: Install latest CUDNN drivers according to Step 3 of Installing CUDA and cuDNN in Ubuntu 22.04 for deep learning

Step 4: Find Compute Capability for the GPU from Your GPU Compute Capability

Step 5: Prepare to compile OpenCV from source code, adopted from Installing OpenCV 4 with CUDA in Ubuntu 22.04

• sudo apt install cmake

• sudo apt install python3-numpy

• sudo apt install libavcodec-dev libavformat-dev libswscale-dev

• sudo apt install libgstreamer-plugins-base1.0-dev libgstreamer1.0-dev

• sudo apt install libgtk-3-dev

• sudo apt install libpng-dev libjpeg-dev libopenexr-dev libtiff-dev libwebp-dev

• sudo apt install git

• git clone https://github.com/opencv/opencv.git

• git clone https://github.com/opencv/opencv_contrib.git

• cd opencv; mkdir build; cd build

• cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local -D WITH_CUDA=ON -D WITH_CUDNN=ON -D WITH_CUBLAS=ON -D WITH_TBB=ON -D OPENCV_DNN_CUDA=ON -D OPENCV_ENABLE_NONFREE=ON -D CUDA_ARCH_BIN={compute capability number in the form of x.x} -D OPENCV_EXTRA_MODULES_PATH=$HOME/opencv_contrib/modules -D BUILD_EXAMPLES=OFF -D HAVE_opencv_python3=ON -D ENABLE_FAST_MATH=1 -D cuda_toolkit_root_dir=/usr/local/cuda-{xx.x} -D CUDNN_INCLUDE_DIR=/usr/include/ -D CUDNN_LIBRARY=/usr/lib/x86_64-linux-gnu/libcudnn.so.8 ..

• make -j {number of CPU cores}

• sudo make install

• sudo ldconfig

Step 6: Create Python 3.10 Virtual Environment using pipx

• sudo apt install python3-pip

• python3 -m pip install --user pipx

• python3 -m pipx ensurepath

• Close and reopen terminal

• pipx install virtualenv

• virtualenv --python=python3.10 {name of environment}

• sudo ln -s /usr/local/lib/python3.10/dist-packages/cv2 {environment path and name}/lib/python3.10/site-packages/cv2

Step 7: Test installation

• Open Python interpreter: python3

• import numpy, cv2

• The following command should indicate that the GPU was located: cv2.cuda.printCudaDeviceInfo(0)