README

This file contain all information required for running the License Plate Detector as part of the COMP 4102 class. We have included two separate testing methods.

Included Files

• app folder
  • contains all required document for running the fastapi application
  • main.py: python file for running fastapi application
• local_run.py: python file for running manual application
• plate_vision.ipynb: Jupyter notebook containing our entire workflow. This includes everything from data processing to model training and model predicting. This notebook is set to read-me as many functions will throw errors due to the missing dataset
• examples folder:
  • contains a series of images for testing purpose.
• report.pdf: our final report
• demo video

Running Instructions It is recommended to install the dependencies manually, requirements.txt doesn't seem to work consistenly. For EasyOCR you may need to install PyTorch separately, you might need a command from their installation guide.

General installation steps:

• (Optional but recommended) In the root directory of the project, create a virtual environment:

python -m venv env

• Activate the environment for the current shell:

env/Scripts/activate

• Install the requirements:

pip install --force-reinstall tensorflow=="2.9.1" easyocr opencv-python matplotlib fastapi uvicorn

• IMPORTANT Download the model from the drive and copy it INTO THE APP FOLDER. If the model is not loaded, then both methods will throw out an error. Due to the model's size, we did not include it into our submission.
• Running the application

  • Method 1 [Faster]: Website

    • Navigate to the app folder. Run the python file main.py with the command: python3 main.py
    • This will open a link in terminal. Follow it and it will lead you to the running website

  • Method 2 [Slower but Easier] : Local file

    • In the root directory is the file local_run.py
    • This python file requires one argument, which is a path to the image you wish to input into the program
    • The program can be run with: python3 local_run.py filepath

Final Note : Per request, we can provide our entire dataset and/or training and processing files