Kinmen, one of Taiwan's outlying islands, aimed to improve public safety and infrastructure in remote areas. With limited power and network coverage, real-time monitoring was difficult. The project, led by Yian Electronics, began with NVIDIA^® Jetson Xavier^™ NX for license plate recognition and basic object detection. As coverage expanded, the system reached its performance limits. In 2025, Yian upgraded to NVIDIA Jetson Orin Nano™ 8G Super Mode to meet higher computing needs. Each unit connects to three full-HD cameras and runs the YOLOv7 model for vehicle tracking, object detection, and behavior alerts. Smart streetlights and sensors respond to events and adjust lighting automatically.

All processing takes place at the edge, which reduces cloud dependency and improves response time. The system's compact, low-power design allowed deployment without major infrastructure changes. It now serves as a stable, scalable foundation for smart infrastructure in remote environments.