河南农业科学 ›› 2023, Vol. 52 ›› Issue (11): 157-166.DOI: 10.15933/j.cnki.1004-3268.2023.11.017

所属专题: 病虫害识别专题

• 农业信息与工程·农产品加工 • 上一篇    下一篇

基于YOLOv3 改进算法的烟叶原料烟草甲识别方法研究

张卫正1,陈赛越扬1,王艳玲1,帖金鑫2,丁佳3,李萌4,李灿林1,苏晓珂1,甘勇1   

  1. (1.郑州轻工业大学计算机科学与技术学院,河南 郑州 450001;2.浙江中烟工业有限责任公司宁波卷烟厂,浙江 宁波 315040;3.河北中烟有限责任公司,河北 石家庄 050051;4.郑州轻工业大学 烟草科学与工程学院,河南 郑州 450001)
  • 收稿日期:2023-06-03 出版日期:2023-11-15 发布日期:2023-12-05
  • 通讯作者: 帖金鑫(1988-),男,山西应县人,工程师,硕士,主要从事卷烟加工工艺研究。E-mail:tiejinxin@zjtobacco.com 李萌(1982-),男,河南长葛人,讲师,博士,主要从事卷烟加工工艺研究。E-mail:limengjeff@126.com
  • 作者简介:张卫正(1982-),男,河南滑县人,讲师,博士,主要从事烟草信息学、数字农业与农业物联网研究。E-mail:weizheng008@vip.126.com
  • 基金资助:
    河南省科技攻关项目(222102210037);河南省高等教育教学改革研究与实践重点项目(2021SJGLX189);浙江中烟工业有限责任公司科技项目(ZJZY2023D020)

Identification of Lasioderma serricorne in Tobacco Leaf Raw Materials Based on Improved YOLOv3 Algorithm

ZHANG Weizheng1,CHEN Saiyueyang1,WANG Yanling1,TIE Jinxin2,DING Jia3,LI Meng4,LI Canlin1,SU Xiaoke1,GAN Yong1   

  1. (1.College of Computer Science and Technology,Zhengzhou University of Light Industry,Zhengzhou 450001,China;2.Ningbo Cigarette Factory,China Tobacco Zhejiang Industrial Co.,Ltd.,Ningbo 315040,China;3.China Tobacco Hebei Industrial Co.,Ltd.,Shijiazhuang 050051,China;4.College of Tobacco Science and Engineering,Zhengzhou University of Light Industry,Zhengzhou 450001,China)
  • Received:2023-06-03 Published:2023-11-15 Online:2023-12-05

摘要: 为了解决烟叶原料中烟草甲人工识别效率低、不准确的问题,实现烟草甲的精准、智能识别,提出一种基于YOLOv3改进算法的烟草甲识别模型。首先采用Random Mix数据增强技术扩充烟草甲数据集;然后,通过引入K-means++算法重新聚类锚框,改进YOLOv3对于烟草甲的识别能力;使用SIoU Loss改进YOLOv3的边界框损失函数,提高模型定位的准确性,加速模型收敛;最后加入特征细化模块过滤冲突信息,优化模型对烟草甲数据集中小目标识别的精准度。结果表明,改进的YOLOv3网络模型对测试集中烟草甲的平均检测精确率、召回率、F1分数和平均精度均值(mAP)分别达到93.26%、88.25%、0.90和94.59%,分别比现有的YOLOv3算法提高了12.21百分点、11.79百分点、0.12、12.40百分点,可为烟草甲的有效识别提供支撑。

关键词: 烟叶, 烟草甲, YOLOv3改进算法, 数据增强, 损失函数, 目标识别

Abstract: In order to solve the problem of low efficiency and inaccuracy in manual recognition of tobacco beetles(Lasioderma serricorne)in tobacco leaf raw materials,and achieve accurate and intelligent recognition of tobacco beetles,a tobacco beetle recognition model was proposed based on the improved YOLOv3 algorithm.Firstly,Random Mix data augmentation technology was used to expand the tobacco beetle dataset;By introducing the K⁃means++ algorithm to recluster anchor boxes,YOLOv3’s recognition ability for tobacco beetles was improved;SIoU Loss was used to improve the Loss function of YOLOv3 bounding box to improve the accuracy of model positioning and accelerate model convergence;Finally,a feature refinement module was added to filter conflict information and optimize the accuracy of the model in identifying small targets in the tobacco beetle dataset.The results showed that,the improved YOLOv3 network model achieved average detection accuracy,recall,F1 score,and mAP of 93.26%,88.25%,0.90,and 94.59% for tobacco beetles in the test set,respectively,increased by 12.21 percentage points,11.79 percentage points,0.12,and 12.40 percentage points compared to the existing YOLOv3 algorithm,providing support for the effective recognition of tobacco beetles.

Key words: Tobacco leaves, Lasioderma serricorne, Improved YOLOv3 algorithm, Data augmentation, Loss function, Target recognition

中图分类号: