林业工程学报2025，Vol.10Issue(3)：130-137,8.DOI:10.13360/j.issn.2096-1359.202311031
基于无人机影像的脐橙树株数识别和树冠提取
Identification of orange tree counts and canopy extraction based on drone imagery
摘要
Abstract
Extracting parameters from navel orange orchards aids in improving the efficiency of tree planting and management,providing more accurate agricultural yield predictions.However,the complex terrain of orchards makes the planting and management of fruit trees challenging.Therefore,more convenient and automated methods are needed to achieve efficient orchard monitoring and management.With the advancements in deep learning and drone technology,the combination of these technologies has greatly enhanced the efficiency and accuracy of fruit tree information extraction,offering significant support for orchard management and decision-making.Traditional manual survey methods for extracting fruit tree parameters are often time-consuming and labor-intensive,resulting in low efficiency.This study employed the YOLOv8 deep learning algorithm combined with drone imagery data from a navel orange orchard in southern Jiangxi Province,China,to automatically extract the number of navel orange trees and their canopy sizes.Experimental results showed that the YOLOv8 model achieved a recognition accuracy of 95.7%and a recall rate of 83.4%for navel orange trees under complex conditions.In terms of canopy width extraction,compared to the measured values,the root mean square error was 0.17,with an average relative error of 4.1%.In this study,the proposed method significantly reduced the time and labor required,compared to that from the traditional manual surveys,providing practical technical support and reference for the management of navel orange orchards.The high accuracy and efficiency of the YOLOv8 model in identifying and measuring tree parameters under complex orchard conditions demonstrated its potential as a valuable tool for modern agricultural practices.By leveraging drone imagery and deep learning,this research highlighted the potential for significant improvements in the monitoring and management of fruit tree orchards,contributing to more precise agricultural practices and better yield predictions.The integration of advanced technologies such as deep learning and drones not only streamlines the process of data collection and analysis,but also enhances the overall management strategies for navel orange orchards.This approach aligned with the growing trend of utilizing automation and intelligent systems in agriculture to break through the limitations of traditional methods.As a result,this study provides a solid foundation for further research and development in automated orchard management,paving the way for more innovative solutions in the agricultural sector.关键词
深度学习/脐橙/树冠提取/株数/YOLOv8Key words
deep learning/navel orange/canopy extraction/tree count/YOLOv8分类
信息技术与安全科学引用本文复制引用
肖志成,况润元..基于无人机影像的脐橙树株数识别和树冠提取[J].林业工程学报,2025,10(3):130-137,8.基金项目
江西省水利科学院研究基金(2022SKTR01). （2022SKTR01）
