Virtual Booth

Artificial Intelligence of Things (AIoT) supports a surveillance-based smart lighting system that operates autonomously in diverse and low-light environments by integrating artificial intelligence with real-time Internet of Things sensing. High-resolution Internet Protocol cameras embedded in smart streetlights continuously capture video streams, which are transmitted through 5G evolution networks to a centralized artificial intelligence server. The server employs a CNN-YOLO object detection framework that processes each frame in a single pass to identify individuals by generating bounding boxes and class labels, enabling rapid and accurate detection in crowded or complex scenes while avoiding multistage processing. Individuals detected are counted automatically, and aggregated counts provide real-time crowd density information. In security applications, the detection of a human within four feet of a smart streetlight initiates an immediate response by increasing lighting brightness to its maximum level, activating an audible alarm, and triggering a Large Language Model (LLM) driven module to select or adapt a prerecorded voice warning based on contextual factors such as time, location, and crowd density. This adaptive audio output enhances deterrence by delivering context-relevant messages while alerting nearby security personnel. The operational process consists of four stages: video capture through camera-equipped smart lights; object detection and counting by means of the CNN-YOLO framework; automated lighting, LLM-based audio, and security responses when predefined proximity thresholds are reached; and visualization and control through a cloud-based dashboard. The dashboard displays detection events along with crowd density metrics and system status, enabling remote monitoring and timely decision-making. The integration of rapid visual detection, context-aware audio alerts, automated security responses, and adaptive lighting control improves safety, operational responsiveness, and situational awareness in public spaces while maintaining high detection accuracy and low latency.