基于改进YOLOv8s的茶叶病虫害检测

doi:10.3969/j.issn.1004-1524.20240836

浙江农业学报 ›› 2025, Vol. 37 ›› Issue (9): 1933-1942.DOI: 10.3969/j.issn.1004-1524.20240836

基于改进YOLOv8s的茶叶病虫害检测

刘睿¹(), 王丽娟²^,^*(), 王秋皓¹, 林旭东¹, 郭启航¹, 许多霖¹, 李文妍¹

¹ 广东工业大学国际教育学院, 广东广州 510006
² 广东工业大学计算机学院, 广东广州 510006

收稿日期:2024-09-13 出版日期:2025-09-25 发布日期:2025-10-15
作者简介:*王丽娟,E-mail:1jwang@gdut.edu.cn
刘睿(2003—),男,广东广州人,学士,主要从事深度学习、计算机视觉研究。E-mail:LR030107@163.com
通讯作者: 王丽娟
基金资助:
广东省自然科学基金(2021A1515012556)

Detection of pest and disease in tea based on improved YOLOv8s

LIU Rui¹(), WANG Lijuan²^,^*(), WANG Qiuhao¹, LIN Xudong¹, GUO Qihang¹, XU Duolin¹, LI Wenyan¹

¹ College of International Education, Guangdong University of Technology, Guangzhou 510006, China
² The School of Computer Science and Technology, Guangdong University of Technology, Guangzhou 510006, China

Received:2024-09-13 Online:2025-09-25 Published:2025-10-15
Contact: WANG Lijuan

摘要/Abstract

摘要：

茶叶种植中病虫害检测主要依靠传统的人工方法,存在效率低、准确性差的问题,本文提出一种改进的YOLOv8s模型,旨在实现茶叶病虫害的实时、高效检测。该模型构建了一个包含5类常见病虫害的数据集,通过引入卷积块注意力模块(CBAM),增强模型的表征能力,提升重要特征提取能力,提高了模型的识别精度;将FasterNet作为骨干网络,在保持高精度的同时缩短检测时间,降低模型复杂度;采用WIoU损失函数,有效解决了在图像处理过程中物体重叠的问题,提高了模型在实际情境下的检测性能。研究结果表明,与传统的YOLOv8s模型相比,改进后的YOLOv8s-CFW模型平均精度均值达到98.2%,提升3.7百分点,每秒浮点运算数(FLOPS)和参数量分别减少43.0%和45.2%,检测时间缩短了34.8%。改进模型在精度与轻量化方面均有显著提升,能满足实时检测需求,为茶叶病虫害的自动化检测提供了可靠的理论支持。

关键词: 茶叶, YOLOv8s, 病虫害检测, 轻量化

Abstract:

The detection of pests and diseases in tea cultivation mainly relies on traditional manual methods, which has problems of low efficiency and poor precision. This paper proposed an improved YOLOv8s model, which aimed to achieve real-time and efficient detection of pests and diseases in tea. A dataset containing five common pests and diseases was constructed in this model. By adding the convolutional block attention module (CBAM), the representation ability of model was enhanced, important feature extraction was enhanced, and the recognition precision of model was improved. The FasterNet backbone network was adopted to reduce detection time while maintaining high precision and significantly reduce the model complexity. By adopting the WIoU loss function, the object overlapping problem in the process of image processing was effectively solved and the detection performance of model in real situations was improved. The results showed that compared with the traditional YOLOv8s model, the mean average precision of improved YOLOv8s-CFW model reached 98.2%, which was improved by 3.7 percentage points, the floating-point operations per second (FLOPS) and parameter count were reduced by 43.0% and 45.2%, respectively, and the detection time was reduced by 34.8%.The improved model demonstrates significant improvements in both precision and lightweight, meets the requirements of real-time detection, and provides reliable theoretical support for the automated detection of pests and diseases in tea.

Key words: tea, YOLOv8s, detection of pest and disease, lightweight

中图分类号:

S571.1

刘睿, 王丽娟, 王秋皓, 林旭东, 郭启航, 许多霖, 李文妍. 基于改进YOLOv8s的茶叶病虫害检测[J]. 浙江农业学报, 2025, 37(9): 1933-1942.

LIU Rui, WANG Lijuan, WANG Qiuhao, LIN Xudong, GUO Qihang, XU Duolin, LI Wenyan. Detection of pest and disease in tea based on improved YOLOv8s[J]. Acta Agriculturae Zhejiangensis, 2025, 37(9): 1933-1942.

图/表 10

图1 茶叶病虫害的病斑表型

Fig.1 Lesion phenotype of pests and diseases in tea

图2 改进前的模型结构图 Detect,检测头;Concat,特征图拼接操作;Upsample,上采样操作;C2f,特征提取模块;Conv,卷积操作;SPPF,快速空间金字塔池化层。

Fig.2 Structure diagram of model before improvement Detect, Detection head; Concat, Stitching operation of feature map; Upsample, Upsampling operation; C2f, Feature extraction module; Conv, Convolution operation; SPPF, Spatial pyramid pooling-fast layer.

图3 卷积块注意力模块原理图

Fig.3 Schematic of CBAM

图4 FasterNet结构图 h,输入特征图的高度;w,输入特征图的宽度;c1~c4,第1至第4阶段输出的通道数;l1~l4,第1至第4阶段堆叠的块数。

Fig.4 Structure of FasterNet h,Height of the input feature map;w,Width of the input feature map;c1-c4,Number of channels output in stages 1 to 4;l1-l4,Number of blocks stacked in stages 1 to 4.

图5 IoU原理图

Fig.5 Schematic of IoU

图6 改进后的模型结构图 Backbone,骨干网络;Neck,特征融合层;Head,输出检测头;Concat,特征图拼接操作;Upsample,上采样操作;C2f,特征提取模块;Conv,卷积操作;SPPF,快速空间金字塔池化层。

Fig.6 Structure diagram of model after improvement Backbone,Backbone network;Neck,Feature fusion layer;Head,Output detection head;Concat,Stitching operation of feature map; Upsample, Upsampling operation;C2f,Feature extraction module;Conv,Convolution operation;SPPF,Spatial pyramid pooling-fast layer.

表1 消融试验结果对比

Table 1 Comparison of ablation experiment results

模型 Model	mAP/%	Params/M	FLOPS/10⁹	检测时间 Detection time/ms	模型大小 Model size/MB
YOLOv8s	94.4	11.13	28.4	4.6	22.5
YOLOv8s-CBAM	96.2	11.39	28.7	4.5	23.1
YOLOv8s-FasterNet	96.5	6.08	16.1	2.9	12.4
YOLOv8s-WIoU	95.4	11.13	28.4	4.5	22.5
YOLOv8s-CBAM-FasterNet	96.8	6.10	16.2	3.0	12.4
YOLOv8s-FasterNet-WIoU	96.5	6.08	16.1	3.0	12.4
YOLOv8s-CBAM-WIoU	96.8	11.39	28.7	4.5	23.1
YOLOv8s-CFW	98.2	6.10	16.2	3.0	12.4

表2 模型改进前后对茶叶病虫害的检测精度对比

Table 2 Comparison of detection precision of pests and diseases in tea before and after improving the model %

模型 Model	平均精度Average precision						平均精度均值 mAP
模型 Model	藻斑病 Algal spot	茶轮斑病 Brown blight	云纹叶枯病 Gray blight	健康叶片 Healthy leaf	茶角盲蝽病 Helopeltis	红斑病 Red spot	平均精度均值 mAP
YOLOv8s	95.5	93.7	95.8	95.1	95.2	91.4	94.5
YOLOv8s-CFW	99.1	97.4	99.5	98.9	98.9	95.4	98.2

图7 改进模型前后对茶叶病虫害的检测效果图中数值为置信度。

Fig.7 Detection effect of pests and diseases in tea before and after improving the model The numbers in the figure represent confidence levels.

表3 不同模型的参数

Table 3 Parameters of different models

模型 Model	mAP/%	Params/10⁶	FLOPS/10⁹	检测时间 Detection time/ms	模型大小 Model size/MB
YOLOV3-tiny	94.0	8.68	12.9	2.7	17.4
YOLOv5s	94.2	7.03	15.8	5.9	14.5
YOLOv7	92.9	36.51	103.2	12.5	74.8
YOLOv8s	94.5	11.13	28.4	4.6	22.5
YOLOv9s	93.7	9.60	38.7	4.7	20.3
YOLOv8s-CFW	98.2	6.10	16.2	3.0	12.4

参考文献 17

[1]	陈禹, 吴雪梅, 张珍, 等. 基于改进YOLOv5s的自然环境下茶叶病害识别方法[J]. 农业工程学报, 2023, 39(24): 185-194.
	CHEN Y, WU X M, ZHANG Z, et al. Method for identifying tea diseases in natural environment using improved YOLOv5s[J]. Transactions of the Chinese Society of Agricultural Engineering, 2023, 39(24): 185-194. (in Chinese with English abstract)
[2]	叶荣, 马自飞, 高泉, 等. 基于改进YOLOv5s-ECA-ASFF算法的茶叶病害目标检测[J]. 中国农机化学报, 2024, 45(1): 244-251.
	YE R, MA Z F, GAO Q, et al. Target detection of tea disease based on improved YOLOv5s-ECA-ASFF algorithm[J]. Journal of Chinese Agricultural Mechanization, 2024, 45(1): 244-251. (in Chinese with English abstract)
[3]	杨佳昊, 左昊轩, 黄祺成, 等. 基于YOLO v5s的作物叶片病害检测模型轻量化方法[J]. 农业机械学报, 2023, 54(S1): 222-229.
	YANG J H, ZUO H X, HUANG Q C, et al. Lightweight method for crop leaf disease detection model based on YOLO v5s[J]. Transactions of the Chinese Society for Agricultural Machinery, 2023, 54(S1): 222-229. (in Chinese with English abstract)
[4]	时雷, 杨程凯, 雷镜楷, 等. 基于改进YOLO v8s的小麦小穗赤霉病检测研究[J]. 农业机械学报, 2024, 55(7): 280-289.
	SHI L, YANG C K, LEI J K, et al. Wheat spikelet detection of Fusarium head blight based on improved YOLO v8s[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(7): 280-289. (in Chinese with English abstract)
[5]	李小祥, 张洁, 秦柯贝, 等. 基于改进YOLOv8的轻量级复杂环境苹果叶片病害检测方法[J]. 南京农业大学学报, 2025, 48(3): 734-743.
	LI X X, ZHANG J, QIN K B, et al. A lightweight method for apple leaf disease detection in complex environments based on improved YOLOv8[J]. Journal of Nanjing Agricultural University, 2025, 48(3): 734-743. (in Chinese with English abstract)
[6]	廖娟, 刘凯旋, 杨玉青, 等. 基于RDN-YOLO的自然环境下水稻病害识别模型研究[J]. 农业机械学报, 2024, 55(8): 233-242.
	LIAO J, LIU K X, YANG Y Q, et al. Rice disease recognition in natural environment based on RDN-YOLO[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(8): 233-242. (in Chinese with English abstract)
[7]	赵兴, 岳喜申, 邬欢欢. 基于改进YOLO v5的苹果叶部病害检测[J]. 江苏农业科学, 2024, 52(8): 183-192.
	ZHAO X, YUE X S, WU H H. Detection of apple leaf disease based on improved YOLOv5[J]. Jiangsu Agricultural Sciences, 2024, 52(8): 183-192. (in Chinese with English abstract)
[8]	左昊轩, 黄祺成, 杨佳昊, 等. 基于改进YOLO v5s的作物黄化曲叶病检测方法[J]. 农业机械学报, 2023, 54(S1): 230-238.
	ZUO H X, HUANG Q C, YANG J H, et al. Improved YOLO v5s-based detection method for crop yellow leaf curl virus disease[J]. Transactions of the Chinese Society for Agricultural Machinery, 2023, 54(S1): 230-238. (in Chinese with English abstract)
[9]	晁晓菲, 池敬柯, 张继伟, 等. 基于PSA-YOLO网络的苹果叶片病斑检测[J]. 农业机械学报, 2022, 53(8): 329-336.
	CHAO X F, CHI J K, ZHANG J W, et al. Apple leaf lesion detection based on PSA-YOLO network[J]. Transactions of the Chinese Society for Agricultural Machinery, 2022, 53(8): 329-336. (in Chinese with English abstract)
[10]	韩鑫, 徐衍向, 封润泽, 等. 基于红外热成像和改进YOLO v5的作物病害早期识别[J]. 农业机械学报, 2023, 54(12): 300-307.
	HAN X, XU Y X, FENG R Z, et al. Early identification of crop diseases based on infrared thermography and improved YOLOv5[J]. Transactions of the Chinese Society for Agricultural Machinery, 2023, 54(12): 300-307. (in Chinese with English abstract)
[11]	张林鍹, 巴音塔娜, 曾庆松. 基于StyleGAN2-ADA和改进YOLO v7的葡萄病害检测方法[J]. 农业机械学报, 2024, 55(1):241-252.
	ZHANG L X, BA Y, ZENG Q S. Grape disease detection algorithm based on StyleGAN2-ADA and improved YOLO v7[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(1):241-252. (in Chinese with English abstract)
[12]	孙道宗, 刘欢, 刘锦源, 等. 基于改进YOLOv4模型的茶叶病害识别[J]. 西北农林科技大学学报(自然科学版), 2023, 51(9): 145-154.
	SUN D Z, LIU H, LIU J Y, et al. Recognition of tea diseases based on improved YOLOv4 model[J]. Journal of Northwest A & F University(Natural Science Edition), 2023, 51(9): 145-154. (in Chinese with English abstract)
[13]	范阳旭. 基于不可分小波和改进YOLOv8的交通标志检测算法[J]. 图像与信号处理, 2024, 13(2): 179-189.
	FAN Y X. Traffic sign detection algorithm based on non-separable wavelets and improved YOLOv8[J]. Journal of Image and Signal Processing, 2024, 13(2): 179-189. (in Chinese with English abstract)
[14]	王美华, 吴振鑫, 周祖光. 基于注意力改进CBAM的农作物病虫害细粒度识别研究[J]. 农业机械学报, 2021, 52(4): 239-247.
	WANG M H, WU Z X, ZHOU Z G. Fine-grained identification research of crop pests and diseases based on improved CBAM via attention[J]. Transactions of the Chinese Society for Agricultural Machinery, 2021, 52(4): 239-247. (in Chinese with English abstract)
[15]	CHEN J Q, MA A Q, HUANG L X, et al. Efficient and lightweight grape and picking point synchronous detection model based on key point detection[J]. Computers and Electronics in Agriculture, 2024, 217: 108612.
[16]	FU Y Y, RAN T, XIAO W D, et al. GD-YOLO: an improved convolutional neural network architecture for real-time detection of smoking and phone use behaviors[J]. Digital Signal Processing, 2024, 151: 104554.
[17]	张学立, 贾新春, 王美刚, 等. 安全帽与反光衣的轻量化检测: 改进YOLOv5s的算法[J]. 计算机工程与应用, 2024, 60(1): 104-109.
	ZHANG X L, JIA X C, WANG M G, et al. Lightweight detection of helmets and reflective clothings: improved YOLOv5s algorithm[J]. Computer Engineering and Applications, 2024, 60(1): 104-109. (in Chinese with English abstract)

基于改进YOLOv8s的茶叶病虫害检测

Detection of pest and disease in tea based on improved YOLOv8s

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 17

相关文章 15

编辑推荐

Metrics

本文评价

[1]	朱铭敏, 张国平, 谭建军, 孙玲姣, 朱黎, 焦洁. 基于YOLOv5s的轻量级茶叶嫩芽终端检测模型[J]. 浙江农业学报, 2024, 36(6): 1413-1424.
[2]	李大华, 孔舒, 李栋, 于晓. 基于改进SSD模型的柑橘叶片病害轻量化检测模型[J]. 浙江农业学报, 2024, 36(3): 662-670.
[3]	范为培, 于晓明, 沈凤龙, 王亮, 王星. 基于改进GhostNet V2的轻量化熊蜂图像分类模型[J]. 浙江农业学报, 2024, 36(12): 2832-2845.
[4]	俞国红, 郑航, 叶云翔, 薛向磊, 傅童. 轻便自走式采茶机的设计与试验[J]. 浙江农业学报, 2023, 35(9): 2233-2239.
[5]	张宁, 陶荣浩, 刘佩诗, 胡含秀, 高琳琳, 郭龙, 祝尊友, 马友华. 不同种类有机肥配施化肥对茶叶生长、品质和土壤肥力的影响[J]. 浙江农业学报, 2023, 35(8): 1844-1852.
[6]	邓美华, 高娜, 吴林土, 徐火忠, 洪海清, 朱有为. 浙江省铅污染源解析与茶叶铅污染风险评价[J]. 浙江农业学报, 2023, 35(5): 1123-1131.
[7]	王金凤, 周琦, 吕玉龙, 陈卓梅. 间作景观树种对茶园生态系统与茶叶生产的影响[J]. 浙江农业学报, 2023, 35(3): 523-533.
[8]	张棚, 杨雪妍, 洪晶, 张娅俐, 田晓静, 张福梅, 曹竑, 陈士恩, 马忠仁, 丁功涛, 宋礼, 罗丽. 贵州湄潭茶区土壤-茶叶系统中微量元素富集规律与产地溯源[J]. 浙江农业学报, 2022, 34(2): 378-390.
[9]	王丽芳, 叶良, 谢忠稳, 党向利. 茶叶抗菌肽粗提物的抑菌活性及其对冷却肉保鲜的影响[J]. 浙江农业学报, 2022, 34(10): 2268-2276.
[10]	李为宁, 柏宣丙, 李兵. 滚筒式茶叶提香机结构参数优化[J]. 浙江农业学报, 2020, 32(2): 348-358.
[11]	杨梅, 罗逢健, 周利, 楼正云, 张新忠, 孙荷芝, 王新茹, 陈宗懋. 国内外茶叶农药最大残留限量标准比较分析与建议[J]. 浙江农业学报, 2020, 32(1): 168-175.
[12]	王彦翔, 张艳, 杨成娅, 孟庆龙, 尚静. 基于深度学习的农作物病害图像识别技术进展[J]. 浙江农业学报, 2019, 31(4): 669-676.
[13]	裴伟，王晓林. 基于图像信息的茶叶二维采摘坐标的提取[J]. 浙江农业学报, 2016, 28(3): 522-.
[14]	余澍琼，张吉红，崔俊霞，张慧丽，倪健波. 利用环介导等温扩增技术检测茶叶中掺杂的大豆成分[J]. 浙江农业学报, 2015, 27(8): 1479-.
[15]	闫建伟;牛素贞;尹杰;宋勤飞;何林;*. 茶叶介电特性的研究[J]. , 2012, 24(3): 0-516.