Smart Traffic Monitoring System Using Computer Vision and Edge Computing

Traffic management systems capture tremendous video data and leverage advances in processing to detect monitor traffic incidents. The collected are traditionally forwarded the center (TMC) for in-depth analysis may thus exacerbate network paths TMC. To alleviate such bottlenecks, we propose utilize edge computing by equipping nodes that close cameras with resources (e.g., cloudlets). A cloudlet, limited as compared TMC, provides capabilities. In this paper, focus on two common monitoring tasks, congestion detection, speed a two-tier based model takes into account of both capability cloudlets unstable condition Our solution utilizes algorithms each task, one implemented at other which designed consideration different resources. While TMC strong computation power, quality it receives depends underlying conditions. On hand, processes very high-quality but captures trade-off. We evaluate performance proposed well via test-bed experiments under weather conditions show our hybrid edge-cloud outperforms cloud-only edge-only solutions.