Dataset Of Central Asia vehicle number plate with Annotations

The Vehicle Licence Plate Dataset is an essential tool for different sectors providing important data for law enforcement, traffic management, Urban planning, etc. The ability to identify vehicles accurately and efficiently plays a vital role in ensuring public safety, optimizing transport systems and improving overall safety.

For our sets we have divided it into 5 classes. These classes include numbers in the format
Kazakhstan: Five digits and three letters. First three digits: registration number, three letters - series, last two letters - region

Kyrgyzstan: four letters and four digits. The first two letters - «KG» (early version - «KS»), the third - region code, the fourth - alternating («H», «P», «M» or «K»),

Uzbekistan: three letters and five digits. First two digits - region code, three letters - series, average three digits - registration number

Tajikistan: six digits and two letters. First four digits - registration number, two letters - series, last two digits - region code

Turkmenistan: four digits and three letters. First letter - series, four digits - registration number, Last two letters - region code

The data set includes images taken from different cameras and under different conditions. This includes different lighting conditions, different angles. We collected 500 data, which we divided into 5 classes (KZ, KG, TM, TJ, UZ).The data are presented in jpg type and size 964px width and height 964px. We also carefully annotated images and saved them in a JSON file.

The dataset is partitioned into "Training" and "Validation" subsets, adhering to a 80:20 ratio, thereby ensuring thorough model training and effective evaluation. The purpose of our data collection is to help develop reliable license plate recognition systems capable of accurately detecting and identifying license plates from different countries and environments.