基于人工智能的作物与草坪杂草识别研究进展  

Research Progress of Crop and Lawn Weed Identification Based on Artificial Intelligence

作　　者：方璇 金小俊[1] 陈勇[1] FANG Xuan;JIN Xiao-jun;CHEN Yong(College of Mechanical and Electronic Engineering,Nanjing Forestry University,Nanjing,Jiangsu 210037,China)

机构地区：[1]南京林业大学机械电子工程学院,江苏南京210037

出　　处：《林业机械与木工设备》2022年第10期30-36,共7页Forestry Machinery & Woodworking Equipment

基　　金：国家自然科学基金项目(32072498);江苏省高等学校大学生创新创业训练计划项目(202110298012Z)。

摘　　要：杂草是造成作物减量、增加生产成本和降低经济增益的重要因素之一。传统杂草识别技术识别效率低,难以适应复杂环境。随着人工智能技术特别是深度卷积神经网络(Deep Convolutional Neural Networks, DCNN)的快速发展,其在杂草识别中的应用愈趋广泛,成为了最具发展潜能的识别方法。不同于以往杂草识别方法分类进行研究现状的综述,从研究对象的角度出发,分别对玉米等农作物、蔬菜等经济作物和草坪进行综述,并将识别方法集中于深度学习(Deep Learning, DL)等人工智能领域。研究中包括但不限于图像分割、杂草预处理、检测、定位及分类。查阅文献发现玉米是农作物中主要的研究对象,经济作物中则是生菜与甜菜,草坪研究起步虽晚但发展迅速,在上述作物的杂草识别中深度学习方法较传统识别方法均取得了更高的准确率及精度,更适于复杂环境中的检测。Weed is one of the important factors that reduce crop yields, increase production costs and reduce economic gains.Traditional weed identification technology has low identification efficiency and is difficult to adapt to complex environments.With the rapid development of artificial intelligence technology, especially deep convolutional neural network, its application in weed identification has become more and more extensive, and it has become the identification method with the most development potential.Different from the review of the research status based on the classification of weed identification methods, this paper reviews crops such as corn, economic crops such as vegetables, and lawns from the perspective of the research object, and focuses the identification methods in the field of artificial intelligence such as deep learning.Research includes but is not limited to image segmentation, weed preprocessing, detection, localization and classification.The literature review found that corn is the main research object among crops, and lettuce and sugar beet are among the commercial crops.Although the research on turfgrass started late,it has developed rapidly.In the weed identification of the above crops,the deep learning method has achieved higher than traditional identification methods.The accuracy and precision are more suitable for detection in complex environments.

关 键 词：杂草识别 人工智能 深度学习 草坪杂草 

分 类 号：S451[农业科学—植物保护]