一种应用于云南省外侵物种识别的边缘计算模型  

An edge computing model applied to the identification of invasive species in Yunnan Province

作　　者：罗玲 宋科 王皓 资彩飞 奉伟 杜铭铭 孙仲享 曹志勇[1] LUO Ling;SONG Ke;WANG Hao;ZI Cai-fei;FENG Wei;DU Ming-ming;SUN Zhong-xiang;CAO Zhi-yong(College of Big Data,Yunnan Agricultural University,Kunming 650500,China;College of Plant Protection,Yunnan Agricultural University,Kunming 650500,China;Yunnan Gaochuang Human Resource Service Co.,Ltd.,Kunming 650221,China)

机构地区：[1]云南农业大学大数据学院,昆明650500 [2]云南农业大学植物保护学院,昆明650500 [3]云南高创人才服务有限公司,昆明650221

出　　处：《湖北农业科学》2023年第10期212-217,222,共7页Hubei Agricultural Sciences

基　　金：国家重点研发计划项目(2021YFD1400200);云南省重大科技专项计划项目(202302AE090020)。

摘　　要：基于MobileNet模型迁移对云南省4种主要外侵物种(鬼针草、喀西茄、水花生和紫茎泽兰)图像进行识别,将宽度倍率为1.0和1.4的MobileNet-v2模型分别应用在本研究数据集上进行试验,分析了MobileNet-v2网络模型识别不稳定的原因,通过增加通道注意力机制模块、更新激活函数和压缩网络层数对模型进行改进。结果表明,改进后的MobileNet-v2模型识别准确率达96.8%,模型参数量仅为1535093。改进后的MobileNet-v2模型识别准确率高、模型参数量少,适合部署于边缘端,能更好地应用于云南省外侵物种防治领域。Based on the MobileNet model migration,four main invasive species(Bidens bipinnata,Cassia plant,water peanut,and Eupatorium adenophorum)images in Yunnan Province were identified.The MobileNet-v2 models with width ratios of 1.0 and 1.4 were applied to the dataset in this study for experiments.The reasons for the unstable recognition of the MobileNet-v2 network model were analyzed,and the model was improved by adding channel attention mechanism modules,updating activation functions,and compressing network layers.The results showed that the improved MobileNet-v2 model had a recognition accuracy of 96.8%,and the model parameter quantity was only 15359093.The improved MobileNet-v2 model had high recognition accuracy and fewer model parameters,making it suitable for deployment at the edge,and could be better applied to the field of invasive species control in Yunnan Province.

关 键 词：外侵物种 边缘计算模型 MobileNet-v2模型 云南省 

分 类 号：S451.1[农业科学—植物保护] TP18[自动化与计算机技术—控制理论与控制工程] TP391.4[自动化与计算机技术—控制科学与工程]