Enhancing Smart City Surveillance: Vehicle Number Plate Detection with YOLO NAS

Urbanization and increased traffic have created a real need for reliable smart city monitoring systems. To enhance smart city surveillance capabilities, this paper introduces a novel method that integrates You Only Look Once (YOLO) Neural Architecture Search (NAS) for vehicle license plate recognition. The idea is to leverage the fast and accurate real-time object detection abilities of YOLO. By integrating it with NAS, which further optimizes the model, the design is adapted specifically for the task of vehicle plate recognition. This ensures both high accuracy and efficient use of resources - making it suitable for practical smart city applications. Through extensive experiments on various datasets, the effectiveness of the YOLO plus NAS-based system is evaluated. The results showcase its ability to reliably detect license plates across diverse lighting conditions, driving speeds, and settings. Moreover, the system demonstrates flexibility to account for a city's changing landscape. Its main perks are real-time processing capability, lower false positives, and versatility for different smart city surveillance needs. Achieving 97% accuracy, the combined use of YOLO and NAS presents a viable way to enhance automated surveillance tools for smart cities. This can lead to improved traffic management, law enforcement, and urban safety overall.