Efficient Vehicle Number Plate Detection and Recognition using Convolutional Neural Networks and YOLO

This paper combines the powerful combination of Convolutional Neural Networks (CNN) and You Only Look Once (YOLOv12) object detection system for vehicle number plate detection and recognition. The novel approach is named as Enhanced Vehicle Number Plate Detection and Recognition (EVNPD&R). The proposed approach provides improved performance than existing VNPD&R systems, addressing the most common challenges including dynamic lighting conditions, obstructions, and diverse number plate patterns. First, the system employs a CNN-bound method for the localization of number plate which utilizes the network’s ability to learn the hierarchical complex features. Second, model training will be done by using variety dataset to enable the accurate detection and isolated number plates for different backgrounds and environment conditions. The YOLOv12 object detection algorithm is used to detect and extract the alphanumeric characters from the detected number plates. The proposed method provides robust and real-time VNPD&R solution with an integration of CNN and YOLOv12 as it performs identification and recognition of number plates with minimal processing latency. The ability of the projected method has been evaluated on different datasets with respect to accuracy, speed and adaptability. The discoveries of the proposed method contribute to the advancement of intelligent transport system, traffic management, vehicle tracking and law enforcement.