基于图像识别的农林自动化巡检管理研究  

Research on automatic inspection management of agriculture and forestry based on image recognition

作　　者：尹翰 张园园 张睿 YIN Han;ZHANG Yuanyuan;ZHANG Rui(Xi’an Traffic Engineering Institute,Xi’an 710300,China;Shaanxi Institute of Technology,Xi’an 710300,China)

机构地区：[1]西安交通工程学院,西安710300 [2]陕西国防工业职业技术学院,西安710300

出　　处：《自动化与仪器仪表》2023年第8期190-194,199,共6页Automation & Instrumentation

基　　金：2023年度陕西省职业技术教育学会研究课题《农村电商物流融合发展助推乡村振兴的对策研究》(2023SZX033);陕西省体育局体育科研常规课题《关于碳中和视角下体育旅游产业高质量发展路径探析的研究》(2022329);陕西国防工业职业技术学院2023年度科研计划项目重点课题《全运会视角下西安都市圈旅游轨道交通组织优化研究》(Gfy23-28);陕西国防工业职业技术学院教育教学改革研究项目《“四贯通,五融合”全过程育人模式课堂创新》。

摘　　要：为提高农林巡检的自动化水平,结合相关的硬软件技术和深度学习算法,提出一种基于改进Alexnet网络的农林自动化巡检与识别系统。在该系统中,以STM32为主控芯片,实现巡检小车队农林图像的采集;通过对Alexnet网络的改进,实现病虫害的自动识别。最后,对上述方案进行验证。结果表明,以上方法对病虫害的识别准确率高达80%以上,且通过网页端可清晰显示识别到病虫害。由此得出以上设计方案可行。In order to improve the automation level of agricultural and forestry inspection,an agricultural and forestry automatic inspection and identification system based on improved Alexnet network is proposed by combining relevant hardware and software technology and deep learning algorithm.In this system,STM32 is used as the main control chip to realize the collection of agricultural and forestry images of patrol teams;Through the improvement of Alexnet network,the automatic identification of diseases and pests is realized.Finally,the above scheme is verified.The results show that the recognition accuracy of the above methods for pests and diseases is more than 80%,and the identified pests and diseases can be clearly displayed through the web page.It is concluded that the above design scheme is feasible.

关 键 词：Alexnet网络 深度学习 病虫害识别 自动巡检 图像处理 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]