编译Proto文件后出现ImportError：无法导入string_int_label_map_pb2

Hey, I've run into this exact issue before when working with the TensorFlow Object Detection API—let's walk through the fixes step by step:

1. Double-check where you ran the protoc command #

The command protoc object_detection/protos/*.proto --python_out=. must be executed from the models/research directory. This ensures the generated .py files end up directly in the object_detection/protos folder (right next to their original .proto files). If you ran it elsewhere, the string_int_label_map_pb2.py file is probably in the wrong place.

Head to ~/Documents/imgmlreport/inception/classification_inception/models/research/object_detection/protos and confirm the string_int_label_map_pb2.py exists. If not, cd back to models/research and re-run the protoc command.

2. Add the required paths to your Python environment #

Even if the .py file exists, Python might not be able to find it because the models/research and models/research/slim directories aren't in its default search path. You have two easy fixes here:

Option 1: Set paths temporarily in the terminal #

Before running your script, run these commands (replace the paths with your actual username):

export PYTHONPATH=$PYTHONPATH:/home/your-username/Documents/imgmlreport/inception/classification_inception/models/research
export PYTHONPATH=$PYTHONPATH:/home/your-username/Documents/imgmlreport/inception/classification_inception/models/research/slim

Option 2: Hardcode paths in your script (recommended) #

Add these lines at the very top of your Python script to automatically add the paths every time you run it:

import sys
sys.path.append('/home/your-username/Documents/imgmlreport/inception/classification_inception/models/research')
sys.path.append('/home/your-username/Documents/imgmlreport/inception/classification_inception/models/research/slim')

3. Verify your protoc version is compatible #

Different versions of the Object Detection API require specific protoc versions. For example:

• TensorFlow 1.x works best with protoc 3.0.0–3.3.0
• TensorFlow 2.x supports newer versions (like 3.19.0+)

Run protoc --version to check your current version. If it's too old or new, download the matching version and replace your current installation.

4. Ensure all .proto files were compiled successfully #

Sometimes the wildcard *.proto might miss some files (due to permissions or corrupted files). Check that every .proto file in object_detection/protos has a corresponding .py file. If string_int_label_map.proto is missing its .py counterpart, compile it individually:

protoc object_detection/protos/string_int_label_map.proto --python_out=.

Try these steps one by one—this should resolve the import error!

内容的提问来源于stack exchange，提问作者송준석