中国农机化学报2026，Vol.47Issue(2)：177-187,11.DOI:10.13733/j.jcam.issn.2095-5553.2026.02.024
深度实例分割方法在农作物观测领域的应用
Applications of deep instance segmentation methods in the field of crop observation
摘要
Abstract
Agriculture is an important part of the global economy,and the development of precision agriculture is of great significance to improving crop yield and quality and reducing resource consumption.In recent years,the application of deep learning,especially instance segmentation methods,has made significant progress in the field of crop observation.This article describes the latest research progress in the application of instance segmentation methods in crop observation,covering a variety of application scenarios from crop phenotypic characteristic analysis,crop pest and disease detection,fruit identification and management,to weed identification and management,crop growth monitoring,etc.A brief description of the instance segmentation methods is provided from the aspects of development history,representative methods,etc.Then the practical application of the instance segmentation method is introduced in detail by category and scenario.Finally,the technical challenges it faces are analyzed,especially data set limitations,identification accuracy in complex environments,real-time performance and computing resources,model generalization capabilities,and cross-modal and heterogeneous data processing capabilities.Based on this.Solutions such as enriching and expanding data sets,optimizing lightweight models and algorithms,multimodal data fusion,innovative application of deep learning technology,and integrated application of intelligent agricultural systems are proposed.关键词
实例分割/深度学习/精准农业/作物表型/病害检测Key words
instance segmentation/deep learning/precision agriculture/crop phenotyping/disease detection分类
信息技术与安全科学引用本文复制引用
Li Meng,Zhai Chengcheng,Qiu Quan..深度实例分割方法在农作物观测领域的应用[J].中国农机化学报,2026,47(2):177-187,11.基金项目
国家自然科学基金面上项目(61973040) （61973040）
