首页 > 过刊浏览>2023年第42卷第4期 >262-269. DOI:10.13300/j.cnki.hnlkxb.2023.04.030
PDF HTML阅读 XML下载 导出引用 引用提醒
基于YOLO-ODM的温室草莓成熟度的快速检测
CSTR:
[cstr]
作者:
作者单位:

1.福建农林大学机电工程学院，福州 350002;2.福建省农业信息感知技术重点实验室，福州 350002

作者简介:

陈仁凡，E-mail：332241094@qq.com

通讯作者:

谢知，E-mail： xz@fafu.edu.cn

中图分类号:

TP391.4

基金项目:

福建省自然科学基金项目（2019J01403）


YOLO-ODM based rapid detection of strawberry ripeness in greenhouse
Author:
Affiliation:

1.College of Mechanical and Electronic Engineering，Fujian Agriculture and Forestry University，Fuzhou 350002，China;2.Fujian Province Key Laboratory of Agricultural Information Perception Technology，Fuzhou 350002，China

  • 摘要
    • |
  • 图/表
    • |
  • 访问统计
    • |
  • 参考文献 [19]
    • |
  • 相似文献 [20]
    • | | |
  • 文章评论
    摘要:

    为解决温室环境下草莓果实快速准确识别问题，提出一种基于改进YOLOv5s的草莓成熟度检测方法。在主干中引入Shuffle_Block作为特征提取网络，从而实现模型轻量化。同时，在颈部结构中使用全维度动态卷积模块（omni-dimensional dynamic convolution， ODConv），以提高网络对草莓目标的信息挖掘能力，降低计算量，并进一步实现轻量化。结果显示，改进后的YOLO-ODM（YOLO with ODConv module）模型的平均精度均值达97.4%，模型体积是7.79 Mb，在GPU上的单张平均检测时间仅11 ms，浮点运算量为6.9×109。与原网络相比，轻量化的YOLO-ODM方法在提高检测精度的同时，模型大小缩减43%，浮点运算量降低52%。以上结果表明，该轻量化方法可快速准确地对温室环境下草莓果实的成熟度进行检测，实现草莓的生长状态监测。

    Abstract:

    An improved YOLOv5s-based method for rapidly detecting strawberry ripeness was proposed to solve the problem of rapid and accurate identification of strawberry fruits in greenhouse. The Shuffle_Block was introduced as a feature extraction network in the backbone to lightweight the model.Meanwhile，the omni-dimensional dynamic convolution （ODConv） module was used in the neck structure to enhance the information mining ability of model for strawberry targets，reduce computational complexity，and further achieve lightweight.The results showed that the average precision of the improved YOLO-ODM model reached 97.4%.The model size is 7.79 Mb.The average detection time on the GPU is only 11 ms per image，and the floating-point operations are 6.9×109 FLOPs.Compared with the original network，the lightweighted YOLO-ODM method improved the accuracy of detection while reducing model size by 43% and floating-point operations by 52%.It is indicated that the lightweighted method can rapidly and accurately detect the ripeness of strawberry fruit in greenhouse，monitor the growth status of strawberries.

    表 6 不同模型的检测性能对比Table 6 Test performance comparison table of different models
    表 3 先验框和尺度匹配结果Table 3 The prior bounding box and scale matching results
    表 2 主干网络Table 2 Backbone network
    表 4 不同网络的消融试验结果Table 4 Ablation experiment results
    表 1 草莓成熟度数据集基本信息Table 1 Strawberry ripeness data set basic information
    图1 YOLOv5s网络结构Fig.1 YOLOv5s network structure
    图2 Shuffle_Block结构Fig.2 Shuffle_Block structure diagram
    图3 全维度动态卷积Fig.3 Omni-dimensional dynamic convolution
    图4 YOLO-ODM结构Fig.4 Structure diagram YOLO-ODM
    图5 模型收敛图Fig.5 Model convergence diagram
    图6 YOLO-ODM模型不同场景下的检测效果Fig.6 YOLO-ODM model with different scene detection effects
    表 5 不同轻量级主干网络的测试对比Table 5 Comparison of different lightweight backbone networks
    参考文献
    [1] 刘小刚,范诚,李加念,等.基于卷积神经网络的草莓识别方法[J].农业机械学报,2020,51(2): 237-244.LIU X G,FAN C,LI J N,et al.Identifcation method of strawberry based on convolutional neural network[J].Transactions of the CSAM,2020,51(2):237-244(in Chinese with English abstract).
    [2] ZHAO Y S,GONG L,HUANG Y X,et al.Robust tomato recognition for robotic harvesting using feature images fusion[J/OL].Sensors,2016,16(2):173[2023-02-17].https://pubmed.ncbi.nlm.nih.gov/26840313/.DOI: 10.3390/s16020173.
    [3] 宋凯,任晓哲.基于YCbCr色彩空间的玉米叶部病斑的图像分割[J].农业工程学报,2008,24(S2): 202-205.SONG K,REN X Z.Image segmentation of disease speckle of corn leaf based on YCbCr color space[J].Transactions of the CSAE,2008,24(S2):202-205(in Chinese with English abstract).
    [4] 赵嘉玮.基于SVM算法的目标分类筛选方法研究[J].无线互联科技,2022,19(23): 123-125.ZHAO J W.Research on strawberry classification and screening method based on SVM algorithm[J].Wireless internet technology,2022,19(23):123-125(in Chinese with English abstract).
    [5] RAJ R,COSGUN A,KULI? D.Strawberry water content estimation and ripeness classification using hyperspectral sensing[J/OL].Agronomy,2022,12(2):425[2023-02-17].https://doi.org/10.3390/agronomy12020425.
    [6] IBBA P,TRONSTAD C,MOSCETTI R,et al.Supervised binary classification methods for strawberry ripeness discrimination from bioimpedance data [J].Scientific reports,2021,11(1):1-13.
    [7] SHAO Y Y,WANG Y X,XUAN G T,et al.Assessment of strawberry ripeness using hyperspectral imaging [J].Analytical letters,2020,54(10):1547-1560.
    [8] HE K M,GKIOXARI G,DOLLáR P,et al.Mask R-CNN[C]//2017 IEEE International Conference on Computer Vision Workshops(ICCV Workshops 2017),October 22-29, 2017,Venice, Italy.Newyork: IEEE,2017:2961-2969.
    [9] LIU W,ANGUELOV D,ERHAN D,et al.SSD: Single shot multibox detector[DB/OL].arXiv,2016:1512.02325[2023-02-17].https://doi.org/10.48550/arXiv.1512.02325.
    [10] LIN T Y,GOYAL P,GIRSHICK R,et al.Focal loss for dense object detection[DB/OL].arXiv,2018:1708.02002[2023-02-17].https://doi.org/10.48550/arXiv.1708.02002.
    [11] BOCHKOVSKIY A,WANG C Y,LIAO H Y M.YOLOv4: optimal speed and accuracy of object detection[DB/OL].arXiv,2020:2004.10934[2023-02-17].https://doi.org/10.48550/arXiv.2004.10934.
    [12] FAN Y C,ZHANG S Y,FENG K,et al.Strawberry maturity recognition algorithm combining dark channel enhancement and YOLOv5[J/OL].Sensors,2022,22(2):419[2023-02-17].https://doi.org/10.3390/s22020419.
    [13] SU F,ZHAO Y P,WANG G H,et al.Tomato maturity classification based on SE-YOLOv3-MobileNetV1 network under nature greenhouse environment[J/OL].Agronomy,2022,12(7):1638[2023-02-17].https://doi.org/10.3390/agronomy12071638.
    [14] 黄彤镔,黄河清,李震,等.基于YOLOv5改进模型的柑橘果实识别方法[J].华中农业大学学报,2022,41(4): 170-177.HUANG T B,HUANG H Q,LI Z,et al.Citrus fruit recognition method based on the improved model of YOLOv5[J].Journal of Huazhong Agricultural University,2022,41(4):170-177(in Chinese with English abstract).
    [15] ZHANG X Y,ZHOU X Y,LIN M X,et al.Shufflenet: An extremely efficient convolutional neural network for mobile devices[DB/OL].arXiv,2017:1707.01083[2023-02-17].https://doi.org/10.48550/arXiv.1707.01083.
    [16] MA N N,ZHANG X Y,ZHENG H T,et al.ShufflenNet v2: Practical guidelines for efficient CNN architecture design[DB/OL].arXiv,2018:1807.11164[2023-02-17].https://doi.org/10.48550/arXiv.1807.11164.
    [17] LI C,ZHOU A J,YAO A B.Omni-dimensional dynamic convolution[C/OL]∥ICLR 2022 conference.[S.l.]: ICLR,2022 [2023-02-17].https://openreview.net/forum？id=DmpCfq6Mg39.
    [18] ZHANG Y F,REN W Q,ZHANG Z,et al.Focal and efficient IoU loss for accurate bounding box regression [J].Neurocomputing,2022,506:146-157.
    [19] 张小花,李浩林,李善军,等.基于EfficientDet-D1的草莓快速检测及分类[J].华中农业大学学报,2022,41(6): 262-269.ZHANG X H,LI H L,LI S J,et al.Rapid detection and classification of strawberries based on EfficientDet-D1[J].Journal of Huazhong Agricultural University,2022,41(6):262-269(in Chinese with English abstract).
    引证文献
    网友评论
    网友评论
    分享到微博
    发 布
引用本文

陈仁凡,谢知,林晨.基于YOLO-ODM的温室草莓成熟度的快速检测[J].华中农业大学学报,2023,42(4):262-269

复制
分享
文章指标
  • 点击次数:621
  • 下载次数: 1516
  • HTML阅读次数: 412
  • 引用次数: 0
历史
  • 收稿日期:2023-02-17
  • 在线发布日期: 2023-08-30
文章二维码

联系我们

电话： 027-87287437，027-87287364
邮政编码： 430070
E-mail：hnlkxb@mail.hzau.edu.cn
地址： 武汉市洪山区狮子山街1号

期刊订阅 Email Alert 国际标准刊号：ISSN 1000-2421 国外邮发代号：BM 3816 国内刊号：CN 42-1181/S 国内发行代号：38-120

华中农业大学学报（自然科学版） ® 2025 版权所有   技术支持：北京勤云科技发展有限公司