基于改进YOLOv5s的花生仁检测系统  

Peanut kernel detection system based on improved YOlOv5s

作　　者：刘居林 李德豪 张振豪 员玉良[2] LIU Ju-lin;LI De-hao;ZHANG Zhen-hao;YUAN Yu-liang(Qingdao Jimo District First Vocational Secondary Professional School,Qingdao 266200,Shandong Province,China;Qingdao Agricultural University,College of Mechanical and Electrical Engineering,Qingdao 266109,Shandong Province,China)

机构地区：[1]青岛市即墨区第一职业中等专业学校,山东青岛266200 [2]青岛农业大学机电工程学院,山东青岛266109

出　　处：《信息技术》2024年第9期168-175,185,共9页Information Technology

摘　　要：为提升对花生仁的检测水平,文中设计了基于改进YOLOv5s的花生仁检测系统。在YOLOv5s神经网络模型引入MobileNet V2模块和CBAM注意力机制后,将其部署到检测系统中。由实验结果可知,部署改进后的神经网络检测系统,检测精度达到98.26%,检测速度提升至改进前的4倍,且权重文件减小了10.5 MB。由此可见,该花生仁检测系统能实现对花生仁的快速、准确检测。In order to improve the detection level of peanut kernel,a peanut kernel detection system based on improved YOLOv5s is designed in this paper.After introducing MobileNet V2 module and CBAM attention mechanism into YOLOv5s neural network model,it is deployed to the detection system.According to the experiment results,after deploying the improved neural network detection system,the detection accuracy reaches 98.26%,the detection speed has been increased by 4 times,and the weight file is reduced by 10.5 MB.Thus,it can be seen that the peanut kernel detection system can realize the rapid and accurate detection of peanut kernel.

关 键 词：花生仁检测 模型改进 花生仁数据集 模型训练 最优模型 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]