基于SSD-MobileNet的火情检测预警系统

Fire Detection Early Warning System Based on SSD-Mobile Net

针对森林火灾实时烟雾检测问题,本文提出一种的森林火情检测预警系统,能够适应在人烟稀少且几乎没有网络的野外环境中,在无人机平台通过目标检测的方法实现火灾烟雾检测。系统主要使用轻量级网络SSD-MobileNet对烟雾图像进行训练,最终在移动端上实现移植。针对大多数森林没有网络覆盖的问题,移动端识别到烟雾信息,通过北斗短报文通信方式即可在没有网络的森林中对火灾的发生的位置信息进行回传。该设计在解放人力资源的同时提升了森林火灾预警的时效性。

Aiming at the real-time smoke detection problem of forest fires,this paper proposes a forest fire detection and early warning system.Through the target detection method in the wild environment with few people and few networks,it can adapt to the fire detection in the Unmanned Aerial Vehicle(UAV)platform.The system mainly uses the lightweight network SSD-MobileNet to train the smoke image and finally implement the migration on the mobile terminal.In view of the problem that most forests do not have network coverage,after the mobile terminal recognizing the smoke information,the Beidou short message communication method can return the location of the fire in the forest without the network.This design enhances the timeliness of forest fire warning while liberating human resources.