基于YOLO框架的农田火源自动检测系统  

Automatic Detection System of Farmland Fire Source

作　　者：白玉 马广焜[1] 彭新茗 王楠翔 白清扬 孟峻锋 刘鑫蕊 朱硕闻 Bai Yu;Ma Guangkun;Peng Xinming;Wang Nanxiang;Bai Qingyang;Meng Junfeng;Liu Xinrui;Zhu Shuowen(College of Software,Shenyang University of Technology,Shenyang 110870)

机构地区：[1]沈阳工业大学软件学院,沈阳110870

出　　处：《现代计算机》2022年第19期33-38,共6页Modern Computer

基　　金：辽宁省2021年大学生创新创业训练计划项目(S202110142021)。

摘　　要：农田安全是社会普遍关注的热点话题,农田火灾也是农田安全隐患中的重要因素。为了降低农田中火源所带来的危害,提出了针对智能火源检测的农田火源自动检测系统。虽然针对火灾的检测,已经存在温度、烟雾传感器等检测手段,但是无法保证监测实时性。针对这一问题,提出了基于YOLOv5框架的农田火源自动检测系统。通过小目标检测技术,实时检测农田当中的火源隐患,节约了人工对火源排查的时间。采用最新的YOLOv5算法,此算法有较短的推理时间和训练时间,大大缩短了从火源产生到人工对火源采取措施的间隔时间。此外,本系统提供多种接口,能够与智能农田系统结合,使农田受到更好的保护。可以将数据整合到云端,通过任意设备调用云端的接口,实现系统的多样性,扩大应用空间。Farmland safety is a hot topic of general concern in the society,and farmland fire is also an important factor in the hidden dangers of farmland safety.In order to reduce the harm of fire source in farmland,an automatic fire source detection system for intelligent fire source detection is proposed.Although there are already temperature and smoke sensors for fire detection,the real-time monitoring cannot be guaranteed.To solve this problem,a farmland fire source automatic detection system based on Yolo V5 framework is proposed.Through the small target detection technology,real-time detection of potential fire sources in farmland,saving the time of manual fire source investigation.The latest yolov5 algorithm is adopted,which has shorter reasoning time and training time,and greatly shortens the interval between fire source generation and manual fire source generation measures.In addition,the system provides a variety of interfaces,which can be combined with the intelligent farmland system to better protect the farmland.The data can be integrated into the cloud,and the cloud interface can be called through any device,realizing the diversity of the system and wide application space.

关 键 词：深度学习 小目标检测 YOLOv5 火灾检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] S429[自动化与计算机技术—计算机科学与技术]