Precise Agriculture:Effective Deep Learning Strategies to Detect Pest Insects

作     者：Luca Butera Alberto Ferrante Mauro Jermini Mauro Prevostini Cesare Alippi Luca Butera;Alberto Ferrante;Mauro Jermini;Mauro Prevostini;Cesare Alippi

作者机构：Universitàdella Svizzera italianaFaculty of InformaticsLugano 6900Switzerland AgroscopeResearch Centre CadenazzoCadenazzo 6594Switzerland IEEE 

出 版 物：《IEEE/CAA Journal of Automatica Sinica》 (自动化学报（英文版）)

年 卷 期：2022年第9卷第2期

页      面：246-258页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 080203[工学-机械设计及理论] 09[农学] 0904[农学-植物保护] 0835[工学-软件工程] 0802[工学-机械工程] 090402[农学-农业昆虫与害虫防治] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：partly supported and funded by the Hasler Foundation under the Project“ARPI:Automated Recognition of Pest Insect Images”(20028) 

主　　题：Computer vision machine learning neural network pest insect pest monitoring Popillia japonica precise agriculture 

摘      要：Pest insect monitoring and control is crucial to ensure a safe and profitable crop growth in all plantation types,as well as guarantee food quality and limited use of *** aim at extending traditional monitoring by means of traps,by involving the general public in reporting the presence of insects by using *** includes the largely unexplored problem of detecting insects in images that are taken in noncontrolled ***,pest insects are,in many cases,extremely similar to other species that are ***,computer vision algorithms must not be fooled by these similar insects,not to raise unmotivated *** this work,we study the capabilities of state-of-the-art(SoA)object detection models based on convolutional neural networks(CNN)for the task of detecting beetle-like pest insects on nonhomogeneous images taken outdoors by different ***,we focus on disambiguating a pest insect from similar harmless *** consider not only detection performance of different models,but also required computational *** study aims at providing a baseline model for this kind of *** results show the suitability of current SoA models for this application,highlighting how FasterRCNN with a MobileNetV3 backbone is a particularly good starting point for accuracy and inference execution *** combination provided a mean average precision score of 92.66%that can be considered qualitatively at least as good as the score obtained by other authors that adopted more specific models.