Requirements

If needed, install the requirements file using a conda env.

Model and Dataset Downloads

You can download all necessary models and datasets via the links provided below (Links no longer active, models needs to be retrained):

• Models: Download Models
  • model_1: Trained for license plate detection.
  • model_2: Trained for character detection.
  • model_3: Custom U-Net trained for license plate denoising.
• Datasets: While not required for running the code, datasets can be downloaded if needed.
  • Rodosol ALPR Dataset
  • Vehicle Rear Dataset

Model Setup

Follow these steps to set up each model correctly:

1. License Plate Detection Model (Model 1)

   • Download model_1_trained_for_license_plate_detection from the Google Drive model_1 folder.
   • Place its contents in the ./all_models/model_1 folder within your project directory.

2. Character Detection Model (Model 2)

   • Download model_2_trained_for_character_detection from the Google Drive model_2 folder.
   • Place its contents in the ./all_models/model_2 folder within your project directory.

3. License Plate Denoising Model (Model 3)

   • Download denoising_checkpoint.pth from the Google Drive model_3 folder.
   • Place it in the ./all_models/model_3_custom_UNet_trained_for_license_plate_denoising folder within your project directory.

Description of each project folder

• all_datasets

  • Contains all dataset required for model training and testing

• all_models

  • Contains all folders for the checkpoints and weights for the trained models

• experiments

  • Contains the results from each experiment run

• illustrations

  • Contains illustrations used in the report and project presentation

• utilities

  • contains code for generating datasets, image enhancing techniques, training methods and methods used for experiments.