Description of Invention
This project presents a real-time cigarette detection system using the YOLOv11n model deployed on a Raspberry Pi. Aimed at reducing public smoking in restricted areas, the system achieves 0.892 mAP@50 and 2.30 FPS in video stream detection, demonstrating reliable performance for small object identification. The NCNN model format is used for optimal video inference on edge devices. This low-cost, plug-and-play solution enables real-time monitoring, making it suitable for restaurants, classrooms, and other smoke-free zones.