Dockerfile配置PYTHONPATH及解决模块导入与文件路径错误

Fix the "No module named 'maskrcnn'" Error #

• Create an empty __init__.py file inside your local maskrcnn directory. This tells Python to recognize the folder as an importable module.
• In your Dockerfile, add an environment variable to include your project root in Python's module search path. Assuming your project files are copied to /app (a standard Docker setup), add this line:

  ENV PYTHONPATH="/app:$PYTHONPATH"
  

• Ensure your Dockerfile sets the working directory before copying project files:

  WORKDIR /app
  COPY . .
  

Fix the "mask_rcnn_coco.h5 not found" Error #

Choose one of the following options based on your setup:

1. Copy local model file into Docker
   If you already have mask_rcnn_coco.h5 in your local maskrcnn folder, add this COPY command to your Dockerfile (after setting the working directory):

   COPY ./maskrcnn/mask_rcnn_coco.h5 /app/maskrcnn/mask_rcnn_coco.h5
   

   Update the path in m_rcnn.py to match: /app/maskrcnn/mask_rcnn_coco.h5, or use a relative path like ./mask_rcnn_coco.h5 if the script runs from the maskrcnn directory.

2. Download model during Docker build
   If you don't have the file locally, add a RUN command to download the official pre-trained COCO model directly into the container (ensure wget is installed):

   RUN mkdir -p /app/maskrcnn && wget -O /app/maskrcnn/mask_rcnn_coco.h5 https://github.com/matterport/Mask_RCNN/releases/download/v2.0/mask_rcnn_coco.h5
   

Sample Complete Dockerfile Snippet #

# Use a compatible Python base image
FROM python:3.9-slim

# Set working directory
WORKDIR /app

# Install required dependencies (adjust based on your project needs)
RUN pip install --no-cache-dir tensorflow keras pillow numpy opencv-python

# Configure PYTHONPATH
ENV PYTHONPATH="/app:$PYTHONPATH"

# Copy all project files into the container
COPY . .

# Uncomment below to download model if not present locally
# RUN mkdir -p /app/maskrcnn && wget -O /app/maskrcnn/mask_rcnn_coco.h5 https://github.com/matterport/Mask_RCNN/releases/download/v2.0/mask_rcnn_coco.h5

# Command to start your inference app
CMD ["python", "app.py"]

内容的提问来源于stack exchange，提问作者BodeTech