Intelligent Water Faucet for Real-Time Water Conservation in Smart Cities

Urbanization necessitates efficient water resource management for sustainable expansion. This work offers a prototype for intelligent water faucets, merging embedded technology and machine learning. Our solution turns traditional faucets into intelligent devices, boosting water saving by identifying user activities in real-time. The system uses a Raspberry Pi to take hand photos, connected via Message Queue Telemetry Transport (MQTT) protocol to a GPU-equipped server. This server implements YOLO v7, a precise and speedy object identification algorithm, discriminating between hands with and without soap. Our model earned an average F1 Score of 99% and a mAP@0.5 of 0.991, enabling automatic water flow adjustment to minimize wastage. This clever water-saving technology connects with the environmental aims of smart cities, offering a realistic strategy to handle water conservation concerns.