Windows系统下Python的dlib安装失败求助（已装Anaconda、Cmake）

Hey there, sorry to hear you're stuck trying to install dlib 19.4 via Conda! Let's work through some common fixes that should get this sorted out for you.

• First, update your Conda installation
  Outdated Conda versions often cause weird package resolution or dependency conflicts. Run these commands first to make sure everything's up to date:

  conda update -n base -c defaults conda
  conda update --all
  

  Once that's done, try re-running your original install command:

  conda install -c conda-forge dlib=19.4
  

• Force Conda to prioritize the conda-forge channel
  Sometimes Conda's default channel priority might prevent it from finding the specific dlib version you want. You can temporarily override channels to force it to use conda-forge exclusively for this install:

  conda install -c conda-forge --override-channels dlib=19.4
  

  If you regularly use conda-forge packages, you can also set it as the strict priority channel permanently (this helps avoid similar issues in the future):

  conda config --add channels conda-forge
  conda config --set channel_priority strict
  

• Check Python version compatibility
  dlib 19.4 is an older version, so it might not support newer Python releases (like 3.9+). First, check your current Python version:

  python --version
  

  If your version is too new, create a dedicated virtual environment with a compatible Python version (3.7 or 3.8 usually works for dlib 19.4) and try installing there:

  conda create -n dlib_env python=3.7
  conda activate dlib_env
  conda install -c conda-forge dlib=19.4
  

• Fallback: Install from source with pip
  Since you already have CMake installed, compiling dlib from source via pip is a reliable fallback. Just make sure you have a C++ compiler set up (on Windows, install Visual Studio Build Tools and select the "Desktop development with C++" workload). Then run:

  pip install dlib==19.4
  

  Note: Compiling can take a few minutes, so be patient and keep your network connection stable.

内容的提问来源于stack exchange，提问作者Pavel Vorobyov