Smart Standalone Edge IoT Device for Traffic Volume Counting in Smart Cities

Abstract

Traffic volume counting survey helps to get an analysis of number and class of vehicles passing through a particular road segment over a period. The work proposes design and development of a standalone edge device to obtain count of vehicles on road based on category like car, bus, truck, two wheeler and auto rickshaws. The YOLO v8 model along with Deep Sort algorithm is deployed over Jetson nano proposed as an edge device. An interactive dashboard is designed to obtain the count and class of each vehicle by specifying a time. The deep learning models are trained using custom real-world datasets and further optimized to be deployed on Jetson nano. Thus, Jetson nano serves as an edge IoT device for vehicle counting. The analysis of the proposed model indicates promising results.