Egyptian motorcycle and vehicle license plate recognition

Egyptian motorcycle and vehicle license plate recognition

연구 분야: Artificial Intelligence

학회: Neural Computing and Applications

License plate recognition is a pivotal technology in global vehicle surveillance systems, yet the development of effective systems for Egyptian license plates, particularly for motorcycles, has been constrained by the lack of a comprehensive dataset. Several Egyptian license plate recognition systems have been developed using deep learning models, image processing procedures, morphological operations, and other techniques, focusing exclusively on car license plates and not addressing the unique challenges of motorcycle license plate recognition. This paper presents a novel YOLOv8-based recognition system designed to address this significant research gap. The proposed system efficiently detects and tracks license plates in video streams, applies advanced image enhancement techniques, and leverages optical character recognition (OCR) to accurately extract alphanumeric characters. One of the key contributions of this work is the creation and annotation of a new dataset comprising over 1200 images of motorcycle license plates, specifically tailored for the YOLO format. The system demonstrates superior performance, achieving a 100% detection rate and an 80% recognition accuracy, surpassing existing algorithms in the field. The recognized license plates are systematically stored in a .csv file for future analysis, enhancing the system’s utility in real-world applications. This research sets a new benchmark for Egyptian license plate recognition, particularly in the underexplored domain of motorcycle plates, and provides a valuable dataset that can be utilized for further advancements in this area.