Rocky Linux 9编译带CUDA支持的OpenCV-Python失败求助

问题概述 #

尝试编译支持CUDA的opencv-python以运行SIFT，执行pip wheel . --verbose时出现CMake配置不完整的错误，且无法找到CMake输出文件定位具体问题。

已执行操作 #

• 递归克隆仓库：git clone --recursive https://github.com/opencv/opencv-python.git
• 进入仓库目录：cd opencv-python
• 设置环境变量并执行编译：

  export CMAKE_ARGS="-DWITH_CUDA=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_FAST_MATH=1 -DCUDA_FAST_MATH=1 -DWITH_CUBLAS=1 -DOPENCV_DNN_CUDA=ON"
  export ENABLE_CONTRIB=1
  pip wheel . --verbose
  

错误输出 #

-- Configuring incomplete, errors occurred!
Traceback (most recent call last):
  File "/tmp/pip-build-env-hrc4y4f6/overlay/lib/python3.9/site-packages/skbuild/setuptools_wrap.py", line 664, in setup
    env = cmkr.configure(
  File "/tmp/pip-build-env-hrc4y4f6/overlay/lib/python3.9/site-packages/skbuild/cmaker.py", line 354, in configure
    raise SKBuildError(msg)

An error occurred while configuring with CMake.
  Command:
    /tmp/pip-build-env-hrc4y4f6/overlay/lib64/python3.9/site-packages/cmake/data/bin/cmake /home/grant/chinese_test/opencv-python/opencv -G 'Unix Makefiles' --no-warn-unused-cli -DCMAKE_INSTALL_PREFIX:PATH=/home/grant/chinese_test/opencv-python/_skbuild/linux-x86_64-3.9/cmake-install -DPYTHON_VERSION_STRING:STRING=3.9.18 -DSKBUILD:INTERNAL=TRUE -DCMAKE_MODULE_PATH:PATH=/tmp/pip-build-env-hrc4y4f6/overlay/lib/python3.9/site-packages/skbuild/resources/cmake -DPYTHON_EXECUTABLE:PATH=/usr/bin/python3 -DPYTHON_INCLUDE_DIR:PATH=/usr/include/python3.9 -DPYTHON_LIBRARY:PATH=/usr/lib64/libpython3.9.so -DPython_EXECUTABLE:PATH=/usr/bin/python3 -DPython_ROOT_DIR:PATH=/usr -DPython_FIND_REGISTRY:STRING=NEVER -DPython_INCLUDE_DIR:PATH=/usr/include/python3.9 -DPython_NumPy_INCLUDE_DIRS:PATH=/tmp/pip-build-env-hrc4y4f6/overlay/lib64/python3.9/site-packages/numpy/_core/include -DPython3_EXECUTABLE:PATH=/usr/bin/python3 -DPython3_ROOT_DIR:PATH=/usr -DPython3_FIND_REGISTRY:STRING=NEVER -DPython3_INCLUDE_DIR:PATH=/usr/include/python3.9 -DPython3_NumPy_INCLUDE_DIRS:PATH=/tmp/pip-build-env-hrc4y4f6/overlay/lib64/python3.9/site-packages/numpy/_core/include -DWITH_CUDA=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_FAST_MATH=1 -DCUDA_FAST_MATH=1 -DWITH_CUBLAS=1 -DOPENCV_DNN_CUDA=ON -DPYTHON3_EXECUTABLE=/usr/bin/python3 -DPYTHON_DEFAULT_EXECUTABLE=/usr/bin/python3 -DPYTHON3_INCLUDE_DIR=/usr/include/python3.9 -DPYTHON3_LIBRARY=/usr/lib64/libpython3.9.so -DBUILD_opencv_python3=ON -DBUILD_opencv_python2=OFF -DBUILD_opencv_java=OFF -DOPENCV_PYTHON3_INSTALL_PATH=python -DINSTALL_CREATE_DISTRIB=ON -DBUILD_opencv_apps=OFF -DBUILD_opencv_freetype=OFF -DBUILD_SHARED_LIBS=OFF -DBUILD_TESTS=OFF -DBUILD_PERF_TESTS=OFF -DBUILD_DOCS=OFF -DPYTHON3_LIMITED_API=ON -DBUILD_OPENEXR=ON -DOPENCV_EXTRA_MODULES_PATH=/home/grant/chinese_test/opencv-python/opencv_contrib/modules -DWITH_CUDA=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_FAST_MATH=1 -DCUDA_FAST_MATH=1 -DWITH_CUBLAS=1 -DOPENCV_DNN_CUDA=ON
  Source directory:
    /home/grant/chinese_test/opencv-python/opencv
  Working directory:
    /home/grant/chinese_test/opencv-python/_skbuild/linux-x86_64-3.9/cmake-build
Please see CMake's output for more information.

error: subprocess-exited-with-error

× Building wheel for opencv-contrib-python (pyproject.toml) did not run successfully.
│ exit code: 1
╰─> See above for output.

note: This error originates from a subprocess, and is likely not a problem with pip.
full command: /usr/bin/python3 /home/grant/.local/lib/python3.9/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py build_wheel /tmp/tmpzppm94gi
cwd: /home/grant/chinese_test/opencv-python
Building wheel for opencv-contrib-python (pyproject.toml) ... error
ERROR: Failed building wheel for opencv-contrib-python
Failed to build opencv-contrib-python
ERROR: Failed to build one or more wheels

排查难点 #

已递归搜索目录树，但未找到CMake相关输出文件，无法定位具体配置错误。

排查与解决建议 #

1. 强制生成CMake日志文件
   手动执行CMake命令并指定日志输出，直接在CMake工作目录运行：

   cd /home/grant/chinese_test/opencv-python/_skbuild/linux-x86_64-3.9/cmake-build
   /tmp/pip-build-env-hrc4y4f6/overlay/lib64/python3.9/site-packages/cmake/data/bin/cmake /home/grant/chinese_test/opencv-python/opencv -G 'Unix Makefiles' --no-warn-unused-cli -DCMAKE_INSTALL_PREFIX:PATH=/home/grant/chinese_test/opencv-python/_skbuild/linux-x86_64-3.9/cmake-install -DPYTHON_VERSION_STRING:STRING=3.9.18 -DSKBUILD:INTERNAL=TRUE -DCMAKE_MODULE_PATH:PATH=/tmp/pip-build-env-hrc4y4f6/overlay/lib/python3.9/site-packages/skbuild/resources/cmake -DPYTHON_EXECUTABLE:PATH=/usr/bin/python3 -DPYTHON_INCLUDE_DIR:PATH=/usr/include/python3.9 -DPYTHON_LIBRARY:PATH=/usr/lib64/libpython3.9.so -DPython_EXECUTABLE:PATH=/usr/bin/python3 -DPython_ROOT_DIR:PATH=/usr -DPython_FIND_REGISTRY:STRING=NEVER -DPython_INCLUDE_DIR:PATH=/usr/include/python3.9 -DPython_NumPy_INCLUDE_DIRS:PATH=/tmp/pip-build-env-hrc4y4f6/overlay/lib64/python3.9/site-packages/numpy/_core/include -DPython3_EXECUTABLE:PATH=/usr/bin/python3 -DPython3_ROOT_DIR:PATH=/usr -DPython3_FIND_REGISTRY:STRING=NEVER -DPython3_INCLUDE_DIR:PATH=/usr/include/python3.9 -DPython3_NumPy_INCLUDE_DIRS:PATH=/tmp/pip-build-env-hrc4y4f6/overlay/lib64/python3.9/site-packages/numpy/_core/include -DWITH_CUDA=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_FAST_MATH=1 -DCUDA_FAST_MATH=1 -DWITH_CUBLAS=1 -DOPENCV_DNN_CUDA=ON -DPYTHON3_EXECUTABLE=/usr/bin/python3 -DPYTHON_DEFAULT_EXECUTABLE=/usr/bin/python3 -DPYTHON3_INCLUDE_DIR=/usr/include/python3.9 -DPYTHON3_LIBRARY=/usr/lib64/libpython3.9.so -DBUILD_opencv_python3=ON -DBUILD_opencv_python2=OFF -DBUILD_opencv_java=OFF -DOPENCV_PYTHON3_INSTALL_PATH=python -DINSTALL_CREATE_DISTRIB=ON -DBUILD_opencv_apps=OFF -DBUILD_opencv_freetype=OFF -DBUILD_SHARED_LIBS=OFF -DBUILD_TESTS=OFF -DBUILD_PERF_TESTS=OFF -DBUILD_DOCS=OFF -DPYTHON3_LIMITED_API=ON -DBUILD_OPENEXR=ON -DOPENCV_EXTRA_MODULES_PATH=/home/grant/chinese_test/opencv-python/opencv_contrib/modules -DWITH_CUDA=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_FAST_MATH=1 -DCUDA_FAST_MATH=1 -DWITH_CUBLAS=1 -DOPENCV_DNN_CUDA=ON 2>&1 > cmake_config_log.txt
   

   执行后查看cmake_config_log.txt即可获取完整的CMake配置错误信息。

2. 检查CUDA环境完整性

   • 验证CUDA Toolkit安装：运行nvcc --version确认版本，确保与当前OpenCV分支支持的CUDA版本匹配
   • 检查CUDA环境变量：确认CUDA_HOME、PATH、LD_LIBRARY_PATH已正确配置，示例命令：

     echo $CUDA_HOME
     echo $PATH | grep cuda
     echo $LD_LIBRARY_PATH | grep cuda
     

3. 清理缓存后重新编译

   • 删除现有构建目录：rm -rf _skbuild
   • 重新设置环境变量并执行编译：

     export CMAKE_ARGS="-DWITH_CUDA=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_FAST_MATH=1 -DCUDA_FAST_MATH=1 -DWITH_CUBLAS=1 -DOPENCV_DNN_CUDA=ON"
     export ENABLE_CONTRIB=1
     pip wheel . --verbose
     

4. 检查基础依赖完整性
   确保系统已安装OpenCV编译所需的基础依赖，以Ubuntu/Debian为例：

   sudo apt-get install build-essential cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev
   sudo apt-get install python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libdc1394-22-dev
   

内容的提问来源于stack exchange，提问作者Grant Curell