基于边缘计算的森林防火视频低功耗终端设计与探索

Design and Exploration of a Low-power Forest Fire Prevention Video Terminal Based on Edge Computing

针对北京林区供电供网困难、森林防火视频监控数据存储占用空间大、人工识别率低等问题,运用深度卷积模型,设计了具备实时监测、自动识别、按需回传、低功耗的森林防火视频终端盒子,可在森林防火视频终端自动监测、识别、报警、回传疑似火情等图像,自动过滤掉90%以上重复、无效视频图像,节约存储资源,减少人工成本。

In response to the difficulties of power and network supply in forest areas in Beijing,the large storage space occupied by forest fire video surveillance data,and the low rate of manual identification,a low-power forest fire video terminal box has been designed using Deep Convolutional Neural Networks.It features realtime monitoring,automatic identification,and on-demand back transmission capabilities.This device can automatically monitor,identify,alert,and back transmit images of suspected fires at the forest fire video terminal,automatically filtering out over 90%of repetitive and invalid video images,saving storage resources and reducing manual labor costs.