An Edge-AI Powered Smart Energy Management System using YOLOv8 based Human Detection with Context-Aware Appliance Control

As the demand for energy efficiency in smart environments continues to rise, this research introduces a novel Internet of Things(IoT) energy management system based on cutting-edge real-time human detection through YOLOv8 on edge devices. The system significantly outperforms traditional ESP32-CAM or Passive Infrared (PIR) sensor approaches by attaining high accuracy, anchor-free object detection through a stable edge AI setup. The innovation lies in merging YOLOv8 decoupled head architecture with context-aware control logic enabling dynamic appliance control based on presence but activity relevance. Upon detection of human presence accurately within a given space (cabin or smart office), the system switches on or off relevant appliances (fan, light) via relay modules. The architecture ensures low latency, enhanced environmental adaptability and energy efficiency. The proposed work achieves high detection accuracy, with a mAP@0.5 of 97.1% and low inference time (8.3 ms) and high FPS (121) is suitable for practical applications.