A Privacy-Preserving Edge Computing Solution for Real-Time Passenger Counting at Bus Stops using Overhead Fisheye Camera

Successful transit planning in smart cities requires automated and efficient passenger counts at the bus stop while also respecting the privacy of passengers-a paramount consideration in the age of responsible AI. In this paper, we describe the implementation and development of a real-time passenger counting system at the bus stop on Nvidia Edge devices with limited battery powered by solar panels, limited memory, and without compromising privacy or incurring substantial costs. Numerous studies have developed and applied computer vision people detection techniques, although this has not been applied and optimized explicitly to edge-computing passenger counting at the bus stop. In this regard, we evaluated different object detection models using an adapted dataset from an overhead fisheye lens camera of passengers at a bus stop that we developed to optimize, analyze, and improve accuracy. We find that YOLO outperforms DetectNet and Faster-RCNN. Then, the best object detection model was optimized and deployed into the Nvidia Jetson Device, and the performance and efficiency of the passenger counting system were evaluated. Deployment via Nvidia DeepStream on the edge showcased more than 50% reduction in GPU, CPU, and memory consumption, enhancing efficiency while conserving energy. As a result, we present a more accurate and efficient edge-computing video analytics solution for an ethically responsible passenger counting system at the smart city bus stop.