Traffic Flow Optimization Using Quantum Computing With Deep Learning

Abstract

When working with urban overload and anxiety, adjustment of traffic flows requires the use of modern calculation appliances. To improve traffic monitoring and vehicle classification, this method combines deep learning with a quantum computer. The model evaluation is possible by the real world complexity provided by the Bangalore traffic data set, which captures different types of vehicles at a busy crossing during the high time. To solve obstacle problems, Yolov5 is used to recognize real -time and recognition and Deepsort is used to guarantee reliable tracking with multiple objects. Trained nerve networks with Learning Vector Quantization (LVQ) improves the accuracy of vehicle classification. A quantum optimizer can help with dynamic root planning and overload reduction using the Quantum Fourier Transform (QFT). By incorporating quantum calculation, several object tracking accuracy (MOTA) is improved 16% compared to the traditional Yolov 5-AV framework. This improvement is achieved through identity switch, miss and reduction in false detections. By combining deep learning with quantum calculation, we can evaluate real -time traffic data, classify vehicles efficiently and dynamically replace the routes, which all improves traffic flows. Urban dynamics are greatly improved through the use of quantum seeking adaptation, which in turn reduces overload and increases traffic efficiency. When it comes to vehicle classification in tight traffic, the LVQ classification is the best option, according to the classification accuracy of different algorithms.