复现DELFCV论文遇'delf'模块缺失及GitHub导入文件不可用问题求助

Let's break down fixes for both issues you're facing while reproducing the DELF paper results:

1. Fixing ModuleNotFoundError: No module named 'delf' #

This error pops up because your Python interpreter can't locate the delf module—it doesn't know to look in the parent directory of your current working folder. Since you're operating in /home/user_name/anaconda3/lib/python3.6/site-packages/tensorflow/models/research/delf/delf/python/examples, the core delf module lives one level up, in /home/user_name/anaconda3/lib/python3.6/site-packages/tensorflow/models/research/delf, which isn't in your default Python path.

Try either of these reliable approaches:

• Quick terminal fix (temporary)
  Run this command before executing your script to add the required path to your environment:

  export PYTHONPATH=/home/user_name/anaconda3/lib/python3.6/site-packages/tensorflow/models/research:$PYTHONPATH
  

  If you want this setting to persist across terminal sessions, add that line to your ~/.bashrc file (or ~/.zshrc if you use Zsh) and run source ~/.bashrc to apply it right away.

• Script-level fix (permanent for this script)
  Insert these lines at the very top of your Python experiment script:

  import sys
  sys.path.append('/home/user_name/anaconda3/lib/python3.6/site-packages/tensorflow/models/research')
  

  This tells Python exactly where to find the delf module every time you run the script.

2. Resolving Missing Import Files from GitHub #

If you're unable to access or locate 3 specific import files from the DELF repo, here are the most likely fixes:

• Fetch submodules you might have missed
  The TensorFlow Models repository uses submodules for some dependencies. If you cloned the repo without the --recursive flag, run these commands to pull in all required submodules:

  cd /home/user_name/anaconda3/lib/python3.6/site-packages/tensorflow/models/research
  git submodule update --init --recursive
  

• Download required pre-trained assets
  Some DELF components (like pre-trained models or config files) aren't stored in the repo itself—you need to download them manually. Check the DELF documentation for details on these files, then place them in the directory specified by your script (usually something like delf/data/).
• Update your local repo and verify file paths
  It's possible the repo structure changed slightly since you cloned it. Run git pull in the /home/user_name/anaconda3/lib/python3.6/site-packages/tensorflow/models/research directory to get the latest version. Then double-check that your script's import statements match the actual file names and folder structure in your local copy.

内容的提问来源于stack exchange，提问作者New bie