Hey there, let's walk through how to get opencv-python set up properly in Anaconda, and fix that numpy version mismatch issue you ran into in PyCharm.

第一步：准备Anaconda虚拟环境（关键！避免版本混乱） #

First off, never install packages directly in the base environment—it's a surefire way to get version conflicts. Let's create a clean environment for your computer vision work:

• Open your Anaconda Prompt (or terminal on Linux/macOS)
• Create a new environment (replace cv_env with your preferred name, and pick a Python version between 3.8-3.10—these are most compatible with opencv):

  conda create -n cv_env python=3.9
  

• Activate the environment:

  conda activate cv_env
  

第二步：安装opencv-python及匹配的依赖 #

We'll use the conda-forge channel here because it has up-to-date, pre-built opencv packages that automatically handle dependency versions (like numpy):

1. Add the conda-forge channel (if you haven't already) to prioritize compatible packages:

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

2. Install opencv-python—this will pull in the exact numpy version it needs:

   conda install opencv-python
   

   If you prefer using pip instead, run this in your activated environment:

   pip install opencv-python
   

   Either way, the package manager will resolve numpy compatibility automatically.

第三步：修复PyCharm中的导入问题 & 版本冲突 #

The error you saw is almost certainly because PyCharm was using a different environment (with a mismatched numpy version) than the one where you installed opencv. Here's how to fix it:

1. Open your project in PyCharm, go to File > Settings > Project: [Your Project Name] > Python Interpreter
2. Click the gear icon in the top-right corner, select Add
3. Choose Conda Environment > Existing environment, then browse to the Python executable of your cv_env environment:
   • Windows: C:\Users\[Your Username]\anaconda3\envs\cv_env\python.exe
   • Linux/macOS: ~/anaconda3/envs/cv_env/bin/python
4. Click OK to apply the changes. Now PyCharm will use the environment with the compatible opencv and numpy versions.

如果你已经有一个混乱的环境： #

If you tried installing opencv in an existing environment and have numpy conflicts, clean it up:

• Activate the problematic environment: conda activate your_old_env
• Uninstall both opencv and numpy:

  conda uninstall opencv-python numpy
  

• Reinstall opencv using the steps above—this will pull in the correct numpy version automatically.

验证安装 #

To make sure everything works, open a Python console in PyCharm (or your activated Anaconda environment) and run:

import cv2
print(cv2.__version__)

If it prints the opencv version without errors, you're good to go!

内容的提问来源于stack exchange，提问作者Abhishek Jain