基于机器视觉的花生果腐病等级划分系统  

Design of a machine vision-based peanut pod rot grading system

作　　者：刘宇 李秀坤 徐琼 刘立鹏 邵利敏[1] LIU Yu;LI Xiukun;XU Qiong;LIU Lipeng;SHAO Limin(College of Mechanical and Electrical Engineering,Hebei Agricultural University,Baoding 071001,China;State Key Laboratory of North China Crop Improvement and Regulation/College of Agronomy,Hebei Agricultural University,Baoding 071001,China)

机构地区：[1]河北农业大学机电工程学院,河北保定071001 [2]河北农业大学农学院/华北作物改良与调控国家重点实验室,河北保定071001

出　　处：《河北农业大学学报》2025年第2期101-107,共7页Journal of Hebei Agricultural University

基　　金：国家重点研发计划项目(2023YFD1202800).

摘　　要：花生果腐病严重影响花生产量及品质,针对人工划分花生果腐病腐烂程度费时费力、精确度低等问题,本研究设计了一种基于机器视觉的单株花生果腐病等级划分系统。为了应对花生根茎、叶片的干扰以及花生间的粘连等问题,在YOLOv5特征提取主干网络后添加CBAM注意力机制,以WIoU的v3版本替换损失函数CIoU,并将改进后的模型命名为YOLOv5-CW。该模型对单株花生识别的Precision值达到了92.3%,mAP值为90.6%。使用PySide6搭建出一款具有数据输入、权重选择、目标检测、数据显示、结果查询等功能的交互界面,操作直观、简单,能够提升用户使用效率。基于机器视觉可以快速、准确地对单株花生果腐病进行等级划分,为果腐病防治和筛选具有高抗性优质基因等育种研究提供有效的技术支持。Peanut pod rot significantly affects peanut yield and quality.Aiming at the problems of time-consuming,labor-intensive,and low accuracy in manually grading the peanut pod rot levels,this study proposed a machine vision-based grading system for individual peanut pods.To deal with the interference of peanut roots,stems,and leaves,as well as adhesion between peanuts,CBAM was added to the YOLOv5 feature extraction backbone network,and the loss function CIoU was replaced with the v3 version of WIoU.The improved model was named YOLOv5-CW.The model achieved a Precision value of 92.3%and a mAP value of 90.6%for peanut pod rot identification of individual plant.Additionally,an interactive interface was designed using PySide6,providing various functionalities such as data input,weight selection,object detection,data display,and result querying.The operation is intuitive and simple,which can improve the user efficiency.Based on machine vision,the grading of individual peanut pod rot can be quickly and accurately classified,providing effective technical support for breeding research such as disease prevention and screening of high-quality genes relating to high-resistance.

关 键 词：花生 果腐病 机器视觉 等级划分系统 改进YOLOv5 

分 类 号：S24[农业科学—农业电气化与自动化]