openCV 4.3.0 build with CUDA

 

 

 

Introduction

---

  openCV 4.3 version이 release되어 블로그 포스팅을 합니다. 이번에는 이전 블로그의 내용이 길어서 간략한 버전으로 작성하였습니다. 4.2 version 과 동일하게 진행되었으며 자세한 내용이 필요한 경우 이전 포스팅( https://www.vbflash.net/31) 을 참조 해주시기 바랍니다. 

 

• 참고바랍니다. (update : 2020.08.05)
• openCV 4.4.0 Build with CUDA  CUDA,TBB,IPP,MKL,Engin,protobuf,openBLAS
  • https://www.vbflash.net/78

• openCV 4.3.0 Build with CUDA  CUDA,TBB,IPP,MKL,Engin,protobuf,openBLAS
  
  • https://www.vbflash.net/40
• openCV 4.5.0 build with CUDA,TBB,IPP,MKL,Engin,protobuf,openBLAS
  • https://www.vbflash.net/31
• openCV Library manage and setting
  • https://www.vbflash.net/32

 아래의 내용대로 진행 중 오류가 생기는 부분은 댓글 남겨주심면 수정하도록 하겠습니다. 

퓨림노 드림

 

 

 

Download

---

• make,  https://cmake.org/download/
• openCV Library
• 4.3.0 version for Windows,  https://sourceforge.net/projects/opencvlibrary/files/4.3.0/opencv-4.3.0-vc14_vc15.exe/download
• openCV extra Module,  https://github.com/opencv/opencv_contrib
• CUDA toolKit
• https://developer.nvidia.com/cuda-downloads?target_os=Windows&target_arch=x86_64
• cuDNN(로그인 필요)
• https://developer.nvidia.com/rdp/cudnn-download
• TBB,  https://github.com/oneapi-src/oneTBB/releases
• IPP,  https://software.intel.com/en-us/ipp
• openBLAS,  https://sourceforge.net/projects/openblas/files/
• eigen,  http://eigen.tuxfamily.org/index.php?title=Main_Page
• protobuf,  https://github.com/protocolbuffers/protobuf/releases

 

 

순서

---

1. cmake download
2. opencv 4.3.0 download
3. opencv extra module download
4. CUDA toolKit install
5. cuDNN install(copy)
6. TBB, IPP, openBLAS, eigen, protobuf download

 

openCV Library manage and setting of various version

openCV Library manage and setting of various version

openCV Library manage and setting of various version

openCV Library manage and setting of various version

openCV Library manage and setting of various version

openCV 4.3.0 build with CUDA

openCV 4.3.0 build with CUDA

openCV Library manage and setting of various version

openCV Library manage and setting of various version

openCV Library manage and setting of various version

openCV Library manage and setting of various version

 

cmake : check & uncheck

---

 

• Index
• 빨강 : generate 이후 추가항목 발생 건
• 주황 : 자동설정
• 참고 : 참고사항(Memo)

 

 

#check

• BUILD_opencv_world (사용자에 따라 개별 설정 후 build)
• OPENCV_ENABLE_NONFREE
• OPENCV_EXTRA_MODULES_PATH : D:/openCV/extracted/opencv_contrib/modules (path 속 "/" 방향 달라짐 주의)
• WITH_TBB
• TBB_DIR : (미입력)
• TBB_ENV_INCLUDE : D:/openCV/extracted/TBB2020/tbb/include
• TBB_ENV_LIB : D:/openCV/extracted/tbb2020/tbb/lib/intel64/vc14/tbb.lib
• TBB_ENV_LIB_DEBUG : D:/openCV/extracted/tbb2020/tbb/lib/intel64/vc14/tbb_debug.lib
• TBB_VER_FILE : D:/openCV/extracted/tbb2020/tbb/include/tbb/tbb_stddef.h
• WITH_EIGEN
• EIGEN_INCLUDE_PATH : D:/openCV/extracted/eigen
• WITH_CUDA 
• WITH_CUBLAS (자동)
• WITH_CUFFT (자동)
• CUDA_FAST_MATH
• CUDA_TOOLKIT_ROOT_DIR : C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.2 (자동)
• WITH_CUDNN
• cmake 3.16.4 version 에서 Advanced 를 선택해야 아래의 항목이 보임
• CUDNN_LIBRARY : C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.2/lib/x64/cudnn.lib 
• CUDNN_INCLUDE_DIR : C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.2/include (자동)
• CUDNN_VERSION : 7.6.5 (자동)
• PROTOBUF_UPDATE_FILES 
• Protobuf_SRC_ROOT_FOLDER : D:/openCV/extracted/protobuf-3.11.3
• openBLAS
• openBLAS_INCLUDE_DIR : D:/openCV/extracted/OpenBLAS0.3.8
• MKL_WITH_TBB 
• MKL_INCLUDE_DIRS : C:/Program Files (x86)/IntelSWTools/compilers_and_libraries_2020.0.166/windows/mkl/include
• MKL_ROOT_DIR : C:/Program Files (x86)/IntelSWTools/compilers_and_libraries_2020.0.166/windows/mkl
• 이후 generate 수행시, LAPACK library 관련 자동설정

 

#uncheck

• build_package
• build_perf_tests
• build_tests
• build_with_static_crt