Windows下TensorFlow Object Detection API报错：ModuleNotFoundError: No module named 'utils'

I recently ran into a frustrating issue while setting up the TensorFlow Object Detection API on Windows 10 (Python 3.6, TensorFlow 1.4.0): my scripts would only run if they were placed directly in the cloned models repository directory. Move them anywhere else, and I'd hit a ModuleNotFoundError: No module named 'utils'. Turns out this was because I hadn't properly configured the Windows equivalent of the Linux export PYTHONPATH=$PYTHONPATH:pwd:pwd/slim environment variable.

Pre-Deployment Steps I Completed #

• Installed TensorFlow and core tools via pip3
• Opened an administrator Command Prompt and ran:

  pip3 install pillow lxml jupyter matplotlib
  

• Cloned the TensorFlow models repository to C:\Users\cdahms\Documents\models
• Downloaded Google Protobuf v3.4.0 for Windows, extracted it to C:\Program Files\protoc-3.4.0-win32
• Navigated to models\research and ran the Protobuf compilation command:

  "C:\Program Files\protoc-3.4.0-win32\bin\protoc.exe" object_detection/protos/*.proto --python_out=.
  

  Verified all corresponding .py files were generated successfully
• Navigated to the object_detection directory, launched Jupyter Notebook, and confirmed object_detection_tutorial.ipynb ran without issues
• Exported the notebook as object_detection_tutorial.py, which worked fine in the repository directory but failed elsewhere

The Fix: Configuring Windows Environment Variables #

To get Python to recognize the API's module paths globally, follow these steps:

1. Open System Advanced Settings: Right-click "This PC" → "Properties" → "Advanced system settings" → "Environment Variables"
2. In the "System variables" section, click "New":
   • Variable name: PYTHONPATH
   • Variable value: Add these two paths separated by a semicolon:

     C:\Users\cdahms\Documents\models\research;C:\Users\cdahms\Documents\models\research\slim
     

3. Locate the PATH variable in System variables, click "Edit", and append %PYTHONPATH% to the end
4. Restart your computer (critical for the environment variables to take effect)
5. After restarting, you can run object_detection_tutorial.py (or any other API script) from any directory without encountering the utils module error

内容的提问来源于stack exchange，提问作者cdahms