基于YOLO神经网络模型的花生智能精选系统设计
白一睿，方辉*，张泽
（四川大学 机械工程学院，四川 成都 610065）
摘要：针对目前使用机器视觉方法对花生进行精选存在精度和效率较低的问题，本文通过搭建花生精选流水线，并将基于YOLO-V5的神经网络模型引入视觉检测部分，提高花生表面质量检测的准确率和效率。对训练后的模型，采用多类别测试集进行测试，并与不同神经网络模型进行对比。结果显示，在不考虑缺陷类别的情况下，缺陷检测准确率达到99%。发芽、发霉、白斑等难识别缺陷类别的检测准确率分别达到97.5%、98%、98.5%。与传统检测模型Faster R-CNN、YOLO-V3等传统模型相比，准确率和检测速度也更为优秀。通过现场应用验证了所用算法和整体系统的可用性。
关键词：深度神经网络；花生表面缺陷检测；机器视觉；YOLO
中图分类号：TP183 文献标志码：A doi：10.3969/j.issn.1006-0316.2023.02.001
文章编号：1006-0316 (2023) 02-0001-06
Design of Peanut Intelligent Sorting System Based on YOLO Neural Network Model
BAI Yirui，FANG Hui，ZHANG Ze
( School of Mechanical Engineering, Sichuan University, Chengdu 610065, China )
Abstract：In view of the low accuracy and efficiency of peanut sorting through machine vision, this paper builds a peanut selection line and introduces the YOLO-V5 based neural network model into the visual detection part to improve the accuracy and efficiency of peanut surface quality detection. The trained model is tested with multi category test set and compared with different neural network models. The results show that the defect detection accuracy reaches 99% without considering the defect category. The detection accuracy of defects such as germination, mildew and white spots, which are difficult to identify, reaches 97.5%, 98% and 98.5% respectively. Compared with the traditional detection models such as Faster R-CNN、YOLO-V3, the accuracy and detection speed are also improved. The availability of the algorithm and the whole system is verified by field application.
Key words：deep neural network；peanut surface defect detection；machine vision；YOLO———————————————
收稿日期：2022-09-16
基金项目：国家自然科学基金（92060114）；四川大学青岛研究院“8122计划”（21GZ30301）
作者简介：白一睿（1998－），男，山西忻州人，硕士研究生，主要研究方向为机器视觉，Email：693666233@qq.com。*通讯作者：方辉（1973－），男，湖南岳阳人，博士，副教授，主要研究方向为数控加工装备误差分析与补偿、精密加工技术及装备，E-mail：jfh@scu.edu.cn。
- 上一篇：基于混合仿生设计的风力机塔架
设为首页 | 加入收藏 | 免责条款
《机械》杂志版权所有 Copyright©2008-2012 Jixiezazhi.com All Rights Reserved
电话：028-85925070 传真：028-85925073 E-mail：jixie@vip.163.com
地址：四川省成都锦江工业开发区墨香路48号 邮编：610063
Powered by PageAdmin CMS