基于YOLOv8的甘蔗茎节高效检测方法  

Efficient Detection Method for Sugarcane Stem Nodes Based on YOLOv8

作　　者：郑镇辉 张淡然 黄伟华[1,3,4] 韦丽娇 郭昌进[1] 陈思睿 吴海韵[1] ZHENG Zhenhui;ZHANG Danran;HUANG Weihua;WEI Lijiao;GUO Changjin;CHEN Sirui;WU Haiyun(Institute of Agricultural Machinery,Chinese Academy of Tropical Agricultural Sciences,Zhanjiang,Guangdong 524091,China;College of Engineering,South China Agricultural University,Guangzhou,Guangdong 510642,China;Key Laboratory of Agricultural Equipment for Tropical Crops,Ministry of Agriculture and Rural Affairs,Zhanjiang,Guangdong 524091,China;Guangdong Engineering Technology Research Center of Precision Emission Control for Agricultural Particulates,Zhanjiang,Guangdong 524091,China;College of Arts,Guangdong Technology College,Zhaoqing,Guangdong 526100,China)

机构地区：[1]中国热带农业科学院农业机械研究所,广东湛江524091 [2]华南农业大学工程学院,广东广州510642 [3]农业农村部热带作物农业装备重点实验室,广东湛江524091 [4]广东省农业类颗粒体精量排控工程技术研究中心,广东湛江524091 [5]广东理工学院艺术学院,广东肇庆526100

出　　处：《热带作物学报》2024年第10期2223-2231,共9页Chinese Journal of Tropical Crops

基　　金：中央级公益性科研院所基本科研业务费专项(No.1630132024014,No.1630132024012,No.1630132024006)。

摘　　要：甘蔗茎节的精确识别对于智能化切种、种植定位及优化蔗园生产管理流程,进而提升产量与经济效益具有显著价值。然而,现有甘蔗茎节检测方法的性能、模型复杂度以及实时性等方面仍存在不足。为有效解决这一问题,本研究选择采用先进的YOLOv8模型,对结构化场景下的甘蔗茎节进行视觉检测。首先,设计野外甘蔗图像采集试验,对采集的甘蔗图像进行人工标记,并建立图像训练集和测试集;接着,采用YOLOv8网络作为甘蔗茎节检测模型,确定最优超参数组合并进行模型训练;最后,进行野外实际识别试验,验证本方法的有效性和高效性。结果表明:本方法在测试集上的精确率、召回率、mAP、单帧推理耗时以及模型大小分别为0.973、0.958、0.974、19.80 ms和6.30 MB。与Edgeyolo_S_Coco网络和Edgeyolo_Tiny网络相比,YOLOv8_n网络的mAP同比分别提高了1.70%和1.30%,单帧推理耗时同比分别降低了4.71 ms和1.50 ms,模型大小同比分别缩减了33.70 MB和17.50 MB。本研究提出的甘蔗茎节检测网络在检测性能和泛化能力上更具优势,能满足户外环境下对算法精度和模型复杂度的需求,为农业智能化生产中的甘蔗收获与种植提供坚实的技术支撑。The accurate identification of sugarcane stem nodes is of significant value for intelligent seed cutting,planting positioning,optimizing the production management process of sugarcane gardens,and improving yields and economic benefits.However,existing sugarcane stem node detection methods still have shortcomings in terms of performance,model complexity,and real-time performance.In order to effectively solve this problem,this study chose to use the advanced YOLOv8 model to visually detect sugarcane stem nodes in a structured scene.First,a field sugarcane image collection experiment was designed,the collected sugarcane images were manually labeled,and an image training set and a test set were established.Then,the YOLOv8 network was used as the sugarcane stem node detection model to determine the optimal hyperparameter combination and conduct model training.Finally,actual recognition experiments in the field are conducted to verify the effectiveness and efficiency of this method.Experimental results show that the precision,recall,mAP,single-frame inference time and model size of our method on the test set are 0.973,0.958,0.974,19.80 ms and 6.30 MB respectively.Compared with the Edgeyolo_S_Coco network and Edgeyolo_Tiny network,the mAP value of the YOLOv8_n network has increased by 1.70%and 1.30%respectively,the single-frame inference time has been reduced by 4.71 ms and 1.50 ms respectively,and the model size has been reduced by 33.70 MB and 17.50 MB respectively.This method has advantages in detection performance and generalization ability,and can effectively meet the requirements for algorithm accuracy and model complexity in outdoor environments,providing solid technical support for sugarcane harvesting and planting in intelligent agricultural production.

关 键 词：目标检测 甘蔗 茎节检测 YOLOv8 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] S566.1[自动化与计算机技术—计算机科学与技术]