metadata
license: mit
pipeline_tag: object-detection
tags:
- License-Plate-Recognizer
- Yolov8m
- Object detection
License Plate Detection Model using YOLOv8
Model Description
This is a deep learning model for detecting and cropping license plates in images, trained using the YOLOv8 object detection architecture. The model takes an image of a vehicle as input and returns a cropped image of the detected license plate.
Dataset
The model was trained on a dataset of 500 images of vehicles with annotated license plates. The dataset was curated to include a variety of license plate types, angles, and lighting conditions.
Model Training
The model was trained using the YOLOv8 architecture with the following hyperparameters:
- Batch size: 32
- Epochs: 50
- Learning rate: 0.001
- Optimizer: Adam
- Loss function: Mean Average Precision (MAP)
Model Performance
The model achieves the following performance metrics on the validation set:
- mAP (mean Average Precision): 0.92
- AP (Average Precision) for license plates: 0.95
- Recall: 0.93
- Precision: 0.94
Usage
To use this model, you can follow these steps:
- Install the required libraries:
pip install ultralytics - Load the model:
model = torch.hub.load('ultralytics/yolov8', 'custom', path='path/to/model.pt') - Load the input image:
img = cv2.imread('path/to/image.jpg') - Preprocess the input image:
img = cv2.resize(img, (640, 480)) - Run the model:
results = model(img) - Extract the cropped license plate image:
license_plate_img = results.crop[0].cpu().numpy()
Example Code
Here is an example code snippet to get you started:
import cv2
import torch
# Load the model
model = torch.hub.load('ultralytics/yolov8', 'custom', path='path/to/model.pt')
# Load the input image
img = cv2.imread('path/to/image.jpg')
# Preprocess the input image
img = cv2.resize(img, (640, 480))
# Run the model
results = model(img)
# Extract the cropped license plate image
license_plate_img = results.crop[0].cpu().numpy()
cv2.imwrite('license_plate.jpg', license_plate_img)