License Plate Images Generation with Diffusion Models

TL;DR

This paper demonstrates that diffusion models can generate realistic license plate images, which can be used to augment training data and improve license plate recognition accuracy, addressing data scarcity due to privacy regulations.

Contribution

The study validates the use of diffusion models for license plate synthesis and provides a publicly available dataset to facilitate further research in this area.

Findings

Synthetic license plates are realistic and diverse.

Augmenting training data with synthetic images improves recognition accuracy.

Diffusion models are effective for license plate image generation.

Abstract

Despite the evident practical importance of license plate recognition (LPR), corresponding research is limited by the volume of publicly available datasets due to privacy regulations such as the General Data Protection Regulation (GDPR). To address this challenge, synthetic data generation has emerged as a promising approach. In this paper, we propose to synthesize realistic license plates (LPs) using diffusion models, inspired by recent advances in image and video generation. In our experiments a diffusion model was successfully trained on a Ukrainian LP dataset, and 1000 synthetic images were generated for detailed analysis. Through manual classification and annotation of the generated images, we performed a thorough study of the model output, such as success rate, character distributions, and type of failures. Our contributions include experimental validation of the efficacy of diffusion models for LP synthesis, along with insights into the characteristics of the generated data. Furthermore, we have prepared a synthetic dataset consisting of 10,000 LP images, publicly available at https://zenodo.org/doi/10.5281/zenodo.13342102. Conducted experiments empirically confirm the usefulness of synthetic data for the LPR task. Despite the initial performance gap between the model trained with real and synthetic data, the expansion of the training data set with pseudolabeled synthetic data leads to an improvement in LPR accuracy by 3% compared to baseline.