python安装opencv出错 PEP 517

我的 python 版本是 3.6 的，当需要安装 opencv 时，使用pip install opencv-python 命令进行安装时，出现错误，如下所示：

Building Custom Rule C:/Users/liqian/AppData/Local/Temp/pip-install-6jpavinn/opencv-python_4bf9b16580254d7190cb574ec2fc8790/opencv/modules/imgproc/CMakeLists.txt
    imgwarp.avx2.cpp
    resize.avx2.cpp
    accum.avx2.cpp
    bilateral_filter.avx2.cpp
    box_filter.avx2.cpp
    filter.avx2.cpp
    color_hsv.avx2.cpp
    color_rgb.avx2.cpp
    color_yuv.avx2.cpp
    median_blur.avx2.cpp
    morph.avx2.cpp
    smooth.avx2.cpp
    sumpixels.avx2.cpp
    opencv_imgproc_AVX2.vcxproj -> C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790\_skbuild\win-amd64-3.6\cmake-build\modules\imgproc\opencv_imgproc_AVX2.dir\Release\opencv_imgproc_AVX2.lib
  C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\MSBuild\Microsoft\VC\v160\Microsoft.CppBuild.targets(517,5): warning MSB8029: 中间目录或输出目录无法驻留在临时目录下，因为这可能会导致增量生成出现问题。 [C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790\_skbuild\win-amd64-3.6\cmake-build\modules\imgproc\opencv_imgproc_AVX512_SKX.vcxproj]
    Building Custom Rule C:/Users/liqian/AppData/Local/Temp/pip-install-6jpavinn/opencv-python_4bf9b16580254d7190cb574ec2fc8790/opencv/modules/imgproc/CMakeLists.txt
    sumpixels.avx512_skx.cpp
    opencv_imgproc_AVX512_SKX.vcxproj -> C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790\_skbuild\win-amd64-3.6\cmake-build\modules\imgproc\opencv_imgproc_AVX512_SKX.dir\Release\opencv_imgproc_AVX512_SKX.lib
  C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\MSBuild\Microsoft\VC\v160\Microsoft.CppBuild.targets(517,5): warning MSB8029: 中间目录或输出目录无法驻留在临时目录下，因为这可能会导致增量生成出现问题。 [C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790\_skbuild\win-amd64-3.6\cmake-build\modules\imgproc\opencv_imgproc_SSE4_1.vcxproj]
    Building Custom Rule C:/Users/liqian/AppData/Local/Temp/pip-install-6jpavinn/opencv-python_4bf9b16580254d7190cb574ec2fc8790/opencv/modules/imgproc/CMakeLists.txt
    imgwarp.sse4_1.cpp
    resize.sse4_1.cpp
    accum.sse4_1.cpp
    box_filter.sse4_1.cpp
    filter.sse4_1.cpp
    color_hsv.sse4_1.cpp
    color_rgb.sse4_1.cpp
    color_yuv.sse4_1.cpp
    median_blur.sse4_1.cpp
    morph.sse4_1.cpp
    smooth.sse4_1.cpp
    opencv_imgproc_SSE4_1.vcxproj -> C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790\_skbuild\win-amd64-3.6\cmake-build\modules\imgproc\opencv_imgproc_SSE4_1.dir\Release\opencv_imgproc_SSE4_1.lib
  C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\MSBuild\Microsoft\VC\v160\Microsoft.CppBuild.targets(517,5): warning MSB8029: 中间目录或输出目录无法驻留在临时目录下，因为这可能会导致增量生成出现问题。 [C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790\_skbuild\win-amd64-3.6\cmake-build\modules\python_bindings_generator\gen_opencv_python_source.vcxproj]
  C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\MSBuild\Microsoft\VC\v160\Microsoft.CppBuild.targets(382,5): error MSB3491: 未能向文件“x64\Release\gen_opencv_python_source\gen_open.7733C44B.tlog\gen_opencv_python_source.lastbuildstate”写入命令行。路径: x64\Release\gen_opencv_python_source\gen_open.7733C44B.tlog\gen_opencv_python_source.lastbuildstate 超过 OS 最大路径限制。完全限定的文件名必须少于 260 个字符。 [C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790\_skbuild\win-amd64-3.6\cmake-build\modules\python_bindings_generator\gen_opencv_python_source.vcxproj]
  C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\MSBuild\Microsoft\VC\v160\Microsoft.CppBuild.targets(517,5): warning MSB8029: 中间目录或输出目录无法驻留在临时目录下，因为这可能会导致增量生成出现问题。 [C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790\_skbuild\win-amd64-3.6\cmake-build\modules\.firstpass\videoio\opencv_videoio_plugins.vcxproj]
  C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\MSBuild\Microsoft\VC\v160\Microsoft.CppBuild.targets(382,5): error MSB3491: 未能向文件“x64\Release\opencv_videoio_plugins\opencv_v.B9CC6B5E.tlog\opencv_videoio_plugins.lastbuildstate”写入命令行。路径: x64\Release\opencv_videoio_plugins\opencv_v.B9CC6B5E.tlog\opencv_videoio_plugins.lastbuildstate 超过 OS 最大路径限制。完全限定的文件名必须少于 260 个字符。 [C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790\_skbuild\win-amd64-3.6\cmake-build\modules\.firstpass\videoio\opencv_videoio_plugins.vcxproj]
  Traceback (most recent call last):
    File "C:\Users\liqian\AppData\Local\Temp\pip-build-env-onh0v2r3\overlay\Lib\site-packages\skbuild\setuptools_wrap.py", line 642, in setup
      cmkr.make(make_args, install_target=cmake_install_target, env=env)
    File "C:\Users\liqian\AppData\Local\Temp\pip-build-env-onh0v2r3\overlay\Lib\site-packages\skbuild\cmaker.py", line 679, in make
      self.make_impl(clargs=clargs, config=config, source_dir=source_dir, install_target=install_target, env=env)
    File "C:\Users\liqian\AppData\Local\Temp\pip-build-env-onh0v2r3\overlay\Lib\site-packages\skbuild\cmaker.py", line 711, in make_impl
      "An error occurred while building with CMake.\n"

  An error occurred while building with CMake.
    Command:
      'C:\Users\liqian\AppData\Local\Temp\pip-build-env-onh0v2r3\overlay\Lib\site-packages\cmake\data\bin/cmake.exe' --build . --target install --config Release --
    Install target:
      install
    Source directory:
      C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790
    Working directory:
      C:\Users\liqian\AppData\Local\Temp\pip-install-6jpavinn\opencv-python_4bf9b16580254d7190cb574ec2fc8790\_skbuild\win-amd64-3.6\cmake-build
  Please check the install target is valid and see CMake's output for more information.
  ----------------------------------------
  ERROR: Failed building wheel for opencv-python
Failed to build opencv-python
ERROR: Could not build wheels for opencv-python which use PEP 517 and cannot be installed directly

开始以为网络的问题，尝试了几次都没有成功。

第一种方案是使用 conda 命令安装成功了, conda install py-opencv.

测试一下，安装成功。

查了半天资料，发现是python版本比较低，直接安装的话，高版本的opencv和低版本的python难以匹配。

因此，还有一种解决方案

安装较低版本的 opencv：

pip install opencv-python==3.4.9.31

包括 opencv-contrib-python :

pip install opencv-contrib-python==3.4.9.31

这样也是能够安装成功的。

参考链接：https://blog.csdn.net/z704630835/article/details/124564044
https://blog.csdn.net/u013548568/article/details/108885490

日常学习记录，一起交流讨论吧！侵权联系~