DESIGN OF A THREE-STEP FRAMEWORK FOR BENGALI LICENSE PLATE DETECTION AND RECOGNITION

Abstract Automated license plate detection and recognition systems are pioneering innovations in the field of computer vision with broad-reaching applications in traffic management, security, and vehicle tracking. Although several reliable systems for languages such as English, a dedicated license plate recognition system is not available for Bangla. This research aims to develop and evaluate an advanced Bangla license plate detection and recognition system, focusing on the linguistic complexity of the Bengali OCR and its reliability. Using cutting-edge YOLOv8 object detection algorithms to precisely identify and locate vehicles in video streams, EasyOCR with a fuzzy string-matching approach to find license plates on vehicle hulls, and a cutting-edge real-time system, this research presents a three-step plan to solve the problems of Bangla license plate detection and recognition. Our organized three-step plan works well, with an overall accuracy of 75.5%, a license plate detection rate of 99%, an OCR accuracy of 93.47%, and an F1 score of 96.6%. These outstanding results not only underscore the system's effectiveness in real-world scenarios but also represent a significant stride forward in the field of computer vision, particularly in addressing the linguistic complexities of the Bengali OCR.