نوع مقاله : مقاله پژوهشی
عنوان مقاله English
نویسندگان English
Over the years, license plate recognition (LPR) has garnered significant attention from the research community due to its wide range of applications, including traffic control, toll collection, police investigations, and more. License plates are typically designed with specific patterns depending on the country or region of issuance. Interpreting these patterns is crucial for improving character recognition accuracy, especially in complex and unconstrained environments. However, existing LPR methods often manually incorporate these rules and patterns. In contrast, the automatic understanding of the "language" of license plates—comprising unique combinations of letters and digits—can significantly enhance system performance, allowing for the dynamic adaptation of recognition models to the specific character sets used in different regions.In this paper, leveraging recent advancements in transformer-based approaches in natural language processing (NLP) and computer vision, we propose a novel model for license plate recognition. The proposed system integrates a vision model for plate image recognition with a language model designed to learn plate-specific rules and perform multi-stage recognition refinement. This process enhances recognition accuracy and improves the model's adaptability to various plate formats. Evaluations conducted on the IR-LPR, UFPR-ALPR, and AOLP datasets demonstrate the superiority of the proposed approach compared to existing methods.
کلیدواژهها English
|
دقت
|
سال
|
معماری
|
مرجع
|
پایگاهداده
|
|
93.3
|
2022
|
مدل بینایی (CNN)
|
[31]
|
IR-LPR
|
|
94.9
|
2024
|
مدل بینایی (CNN)
|
[35]
|
|
|
18/94
|
2024
|
مدل بینایی (ترنسفورمری)
|
پیشنهادی
|
|
|
97.12
|
2024
|
مدل بینایی + مدل زبانی (ترنسفورمری)
|
پیشنهادی
|
|
|
97.57
|
2019
|
مدل بینایی (CNN)
|
[3]
|
UFPR-ALPR
|
|
96.87
|
2020
|
مدل بینایی (CNN)
|
[36]
|
|
|
90
|
2024
|
مدل بینایی + مدل زبانی (ترنسفورمری)
|
[37]
|
|
|
99.93
|
2024
|
مدل بینایی + مدل زبانی (ترنسفورمری)
|
پیشنهادی
|
|
|
99.2
|
2021
|
مدل بینایی (CNN)
|
[38]
|
AOLP
|
|
98.36
|
2018
|
مدل بینایی (CNN)
|
[39]
|
|
|
99.4
|
2024
|
مدل بینایی + مدل زبانی (ترنسفورمری)
|
پیشنهادی
|