Dark Mode
Global Vehicle License Plate OCR Dataset
Government & Civic Records
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
This dataset provides a substantial collection of over 1.2 million labelled license plates from vehicles worldwide, specifically designed for License Plate Recognition (LPR) and Optical Character Recognition (OCR) tasks. It includes images sourced from various web domains, offering a robust foundation for developing and enhancing automated vehicle identification systems. The dataset serves as a crucial resource for computer vision applications focusing on identifying and extracting information from vehicle registration plates.
Columns
The dataset includes several key variables in its CSV files to facilitate detailed analysis and model training:
- file_name: The filename of the original car photograph.
- license_plate.country: Indicates the country where the vehicle was captured.
- bbox: Provides normalised bounding box coordinates for the car.
- license_plate.visibility: Describes the visibility type of the license plate.
- license_plate.id: A unique identifier for each license plate.
- license_plate.mask: Contains normalised coordinates of the license plate itself.
- license_plate.rows_count: Specifies whether the license plate is single-line or double-line.
- license_plate.number: The recognised text of the license plate.
- license_plate.serial: Applicable only for UAE numbers, this indicates the license plate series.
- license_plate.region: Applicable only for UAE numbers, this denotes the license plate subregion.
- license_plate.color: Applicable only for Saudi Arabia, this refers to the colour of the international plate code.
Distribution
The dataset encompasses over 1.2 million annotated license plates. The full version includes 1,200,000 images and OCR annotations. While specific numbers for rows/records in the example CSV files are not universally provided, the Brazil example includes a CSV file of 16.08 kB, and the total data explored is 111.2 MB. The data files are typically in CSV format.
Usage
This dataset is ideally suited for a variety of applications, including:
- Developing and training Automatic Number Plate Recognition (ANPR) systems.
- Enhancing Optical Character Recognition (OCR) models for text extraction from images.
- Research in vehicle license plate location and detection.
- Creating solutions for traffic management, law enforcement, and smart city initiatives.
- Building models for computer vision tasks related to vehicle identification.
Coverage
The dataset offers global geographic scope, with labelled plates from various countries around the world. Specific regions mentioned include Brazil, Estonia, Finland, Kazakhstan, Lithuania, Serbia, and UAE. There are specific notes regarding data availability for UAE numbers (including serial and region) and Saudi Arabia (for international plate color). The expected update frequency for this dataset is 'Never'.
License
Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)
Who Can Use It
This dataset is intended for a range of users, including:
- Machine Learning Engineers and Data Scientists developing ANPR and OCR models.
- Researchers in computer vision and artificial intelligence.
- Automotive Industry professionals working on intelligent vehicle systems.
- Law enforcement agencies for surveillance and identification systems.
- Academics and Students for educational and research purposes.
Dataset Name Suggestions
- Global Vehicle License Plate OCR Dataset
- Worldwide ANPR Dataset
- Multi-Country License Plate Recognition Data
- Annotated License Plates for Computer Vision
- Universal Vehicle Registration Plate Dataset
Attributes
Original Data Source: Global Vehicle License Plate OCR Dataset
QUALITY