Enabling YOLOv2 Models to Monitor Fire and Smoke Detection Remotely in Smart Infrastructures

This paper presents implementation of a centralized antifire surveillance management system based on video camera. The system provides visualization information and an optimal guide for quick response of fire and smoke detection. We utilize deep learning model (YOLOv2) and Jetson nano board with Raspberry Pi camera as Internet of things (IoT) sensors. The smart cameras will be mounted in indoor and outdoor environments, and connected to the centralized computer via ethernet cables and communication protocols according to an IoT scheme. Specific software will be used in the centralized computer to show video stream from each camera, in real-time while these cameras are responsible for detecting fire and smoke objects and to generate the alarms accordingly. The proposed approach is able to monitor and supervise fire and smoke detection from different cameras remotely. It is suitable for targeted applications such as smart cities, smart transports, or smart infostructures.