Development of Edge Camera System for Vehicle Detection System Using Local AI Optimizer Based on Minimum Network Resource

This paper proposes an edge camera system for a vehicle detection system using AI local optimization method utilizing minimal network transmission data. Currently, various AI CCTVs are installed, but if they are installed in an area without data network support, updates are slow and optimization is difficult. We improve traffic object recognition by remotely optimizing the detector with minimal data in a 3G or so communication environment, and use it to estimate the speed and location of the vehicle. Local AI optimizer utilizes optimized weight data using DBs using environmental data-based background images, and vehicle speed estimation utilizes warping data-based tracking data. We confirmed the high sensing performance and speed recognition rate through certification exam of the proposed edge camera system.