下载opencv和opencv_contrib，cmake过程中要下载的一些包可以手动下载配置，如果网络较好，也可以等待自动下载。主要记录的是cmake命令：
cmake -D CMAKE_BUILD_TYPE=RELEASE \
-D BUILD_opencv_python3=YES \
-D CMAKE_INSTALL_PREFIX=/usr/local/opencv4_5_0 \
-D WITH_TBB=ON \
-D ENABLE_FAST_MATH=1 \
-D CUDA_FAST_MATH=1 \
-D WITH_CUBLAS=1 \
-D WITH_CUDA=ON \
-D BUILD_opencv_cudacodec=On \
-D WITH_CUDNN=ON \
-D CUDA_ARCH_BIN=8.6 \
-D WITH_OPENGL=ON \
-D WITH_OPENCL=ON \
-D WITH_GSTREAMER=ON \
-D OPENCV_GENERATE_PKGCONFIG=ON \
-D OPENCV_PC_FILE_NAME=opencv4.pc \
-D OPENCV_PYTHON3_INSTALL_PATH=/home/user/data2t/software/anaconda3/envs/py38/lib/python3.8/site-packages \
-D OPENCV_ENABLE_NONFREE=ON \
-D CUDA_CUDA_LIBRARY=/usr/local/cuda/lib64/stubs/libcuda.so \
-D CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda \
-D OpenCL_LIBRARY=/usr/local/cuda/lib64/libOpenCL.so \
-D CUDNN_LIBRARY=/usr/local/cuda/lib64/libcudnn.so \
-D OpenCL_INCLUDE_DIR=/usr/local/cuda/include \
-D OPENCV_EXTRA_MODULES_PATH=/home/user/data1t/3rdparty/opencv-4.5.0/opencv_contrib/modules \
-D PYTHON3_LIBRARIES=/home/user/data2t/software/anaconda3/envs/py38/lib/libpython3.8.so \
-D PYTHON3_PACKAGES_PATH=/home/user/data2t/software/anaconda3/envs/py38/lib/python3.8/site-packages \
-D PYTHON3_INCLUDE_PATH=/home/user/data2t/software/anaconda3/envs/py38/include/python3.8 \
-D INSTALL_PYTHON_EXAMPLES=ON \
-D INSTALL_C_EXAMPLES=ON \
-D BUILD_SHARED_LIBS=ON \
-D BUILD_EXAMPLES=ON .. \
-D BUILD_opencv_python2=OFF \
-D BUILD_opencv_python3=ON \
-D HAVE_opencv_cudaarithm=ON \
-D HAVE_opencv_cudawarping=ON \
-D PYTHON3_EXECUTABLE=/home/user/data2t/software/anaconda3/envs/py38/bin/python
make -j 12
sudo make install
这里有个opencv编译参数一览表：https://blog.csdn.net/qq_31112205/article/details/107039695