UK Number Plate Recognition Dataset

Overview

Welcome to the UK Number Plate Recognition Dataset, a comprehensive collection designed for training and evaluating number plate recognition models. This dataset is curated to meet the diverse needs of researchers, developers, and enthusiasts in the field of computer vision and artificial intelligence, specifically on number plate recognision within the UK.

Dataset Contents

1. Image Samples:

• High-resolution images capturing various scenarios and lighting conditions.
• Low-resolution images, to provide scenarios where conditions aren't perfect.
• Diverse backgrounds, including urban, suburban, and rural settings.
• Images obtained from both static and dynamic scenes.

2. Annotations:

• Accurate bounding box annotations for each detected number plate.
• Annotated regions include variations in plate sizes, styles, and orientations.
• Detailed annotations for character recognition within the plates.*

3. Variety of Plates:

• A wide range of license plate designs and colours found within the United Kingdom.
  • Inclusion of foreign plates, as number plates found in the UK aren't strictly UK based.
• Inclusion of standard plates as well as personalized/custom plates.

4. Challenging Conditions:

• Adverse weather conditions such as rain, fog, and varying light intensities.
• Occlusions and overlapping objects to simulate real-world challenges.

5. Data Split:

• Training, validation, and test sets for seamless model development and evaluation.
• Stratified sampling to ensure a balanced distribution of plate characteristics.

Potential Use Cases

• Automatic Number Plate Recognition (ANPR): Train models for accurate identification of number plates in real-time scenarios, using great and not-so-perfect images as well as variety in the plates themselves, such as inclusion of foreign plates.
• Traffic Monitoring and Management: Utilise the dataset to develop systems for traffic analysis and monitoring.
• Security and Surveillance: Enhance security applications by implementing robust number plate recognition systems.

Thank You

• I created this dataset after a spike of interest in ANPR technology and generalised number plate recognision. I hope this may help other people, in the way this is helping myself. Thank you.

                        @misc{
                            numberplate-kzhpx_dataset,
                            title = { NumberPlate Dataset },
                            type = { Open Source Dataset },
                            author = { praveen },
                            howpublished = { \url{ https://universe.roboflow.com/praveen-cvpvo/numberplate-kzhpx } },
                            url = { https://universe.roboflow.com/praveen-cvpvo/numberplate-kzhpx },
                            journal = { Roboflow Universe },
                            publisher = { Roboflow },
                            year = { 2024 },
                            month = { oct },
                            note = { visited on 2024-11-01 },
                            }