Enhancing Operational Efficacy of Smart Parking Facilities through Intelligent License Plate Recognition

Rapid population expansion and urbanization have increased car ownership, raising security concerns and parking issues. Safe parking is in high demand as car use rises, making sustainable urban growth challenging. Recent advances in artificial intelligence have made license plate identification and retrieval significant. Therefore, intelligent license plate identification systems are crucial for smart parking facility management. This innovative system could improve smart parking management and reduce car parking issues. Using automated license plate recognition technology decreases the time spent finding a suitable place for parking. Hence, the primary objective of this research is to introduce a proposed methodology for the recognition of car license plates, with the aim of enhancing the operational efficacy of smart parking facilities. To achieve this objective, two methodologies, namely Inpainting and Super Resolution, were employed in conjunction with the UNet and SRCNN algorithms. The proposed approach was evaluated on a comprehensive dataset comprising 4171 instances of Persian license plates. To evaluate the effectiveness of the models, several performance metrics were employed. The findings revealed that the U-Net model, incorporating a kernel regularizer, exhibited satisfactory performance, achieving an accuracy of 0.79. Similarly, the Super-Resolution Convolutional Neural Network (SRCNN) model, also utilizing a kernel regularizer, demonstrated a commendable accuracy of 0.85. Furthermore, the proposed intelligent approach in this research can be a valuable tool in the realms of inpainting and super resolution, particularly within the domain of smart parking.