Ubuntu18.04+opencv3.4.0+CUDA10+Optical Flow

#安装Opencv(3.4.3与CUDA似乎有版本兼容问题！！！！)
1.下载opencv3.4.0和opencv_contrib3.4.0于opencv-3.4.0文件夹下
2.mkdir build && cd build
3.cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/usr/local -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-3.4.0/modules -DOPENCV_ENABLE_NONFREE=True ..
4.错误

-- Could not find OpenBLAS include. Turning OpenBLAS_FOUND off
-- Could not find OpenBLAS lib. Turning OpenBLAS_FOUND off
-- Could NOT find Atlas (missing: Atlas_CBLAS_INCLUDE_DIR Atlas_CLAPACK_INCLUDE_DIR Atlas_CBLAS_LIBRARY Atlas_BLAS_LIBRARY Atlas_LAPACK_LIBRARY) 
-- A library with BLAS API not found. Please specify library location.
-- LAPACK requires BLAS
-- A library with LAPACK API not found. Please specify library location.
-- VTK is not found. Please set -DVTK_DIR in CMake to VTK build directory, or to VTK install subdirectory with VTKConfig.cmake file

解决：sudo apt-get install libopenblas-dev
5.下载ippicv
于build文件夹，修改opencv-3.4.0/3rdparty/ippicv的47行file:///home/xxx/opencv-3.4.0/build/
6.cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/usr/local -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-3.4.0/modules -DOPENCV_ENABLE_NONFREE=True ..
7.如需，下载face_landmark_model.dat，下载tiny-dnn
8.修改opencv-3.4.0/opencv_contrib-3.4.0/modules/face/CMakeLists.txt的19行file:///home/xxx/opencv-3.4.0/build/
9.make -j8, 遇到错误Built target libprotobuf,重新cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/usr/local -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-3.4.0/modules -DENABLE_CXX11=1 -DOPENCV_ENABLE_NONFREE=True ..
10.由于CUDA和NVIDIA的版本更新，出现dynlink_nvcuvid.h和nvcuvid.h的错误，修改modules/cudacodec/src/precomp.hpp中#if CUDA_VERSION >= 9000 && CUDA_VERSION < 10000，重新cmakecmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/usr/local -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-3.4.0/modules -DWITH_NVCUVID=ON -DBUILD_opencv_cudacodec=OFF -DBUILD_TIFF=ON -DENABLE_CXX11=1 -DOPENCV_ENABLE_NONFREE=True ..
11.make clean, make, sudo make install
#安装Optical_Flow
12.下载Optical_Flow

mkdir -p build
cd build
cmake ..
make

将需要转换的视频文件夹test移动到build文件夹中

./compute_flow --gpuID=0 --type=1 --skip=1 --vid_path=test --out_path=test_flow