基于U-Net网络的果园垄间导航线检测方法综述OA

With the development of orchard scale,standardization,close planting and dwarfing in China,the demand for the automation level of intelligent agricultural machinery in plant protection,picking and other links is increasingly urgent.Autonomous navigation has become the key technology for the intelligent upgrading of orchard agricultural machinery.Orchard ridge is an unstructured environment,and the application of traditional navigation methods is limited by strong interference factors such as variable lighting and shading of branches and leaves.Machine vision has the advantages of low cost and strong environmental adaptability,and is the mainstream technology route of orchard ridge navigation.This paper reviews the research progress and related methods of orchard ridge navigation line detection based on U-Net network in recent years.This paper first introduces the core evaluation indexes of semantic segmentation model and the method of data set production,and focuses on the semantic segmentation method based on U-Net and its improvement strategies,and summarizes the key technologies and application scenarios of road edge determination and navigation line fitting between ridges in detail.Finally,the shortcomings of the current research are analyzed and the future development direction is prospected,which provides reference for the further optimization of visual navigation technology of orchard intelligent agricultural machinery.