This repo contains a PyTorch finetuning script for a Mask R-CNN model with a ResNet-50-FPN backbone (from the Mask R-CNN paper) on a large dataset of license plates.
pip install -r requirements.txt
- Download dataset
- Clean dataset using
dataset_scripts/clean_dataset.py python train.py
python inference.py <path to model pickle file> <path to image>
Original Image:
Detection:
Original Image:
Detection:
Original Image:
Detection:
This code was adapted from this tutorial.