基于动态自适应特征网络的草莓叶片病害检测算法

doi:10.3969/j.issn.1004-1524.20250271

浙江农业学报 ›› 2026, Vol. 38 ›› Issue (8): 1660-1673.DOI: 10.3969/j.issn.1004-1524.20250271

基于动态自适应特征网络的草莓叶片病害检测算法

余风军¹(), 周晓平²^,^*(), 龚谭飞³

¹ 河南司法警官职业学院信息技术系, 河南郑州 450046
² 郑州大学电气与信息工程学院, 河南郑州 450001
³ 河南农业大学信息与管理科学学院, 河南郑州 450046

收稿日期:2025-04-01 出版日期:2026-08-25 发布日期:2026-09-14
作者简介:余风军,研究方向为计算机信息技术。E-mail: yufengjun9888@163.com
通讯作者: *周晓平,E-mail: zhouxiaoping6603@163.com
基金资助:
国家自然科学基金(32371633);河南省科技攻关项目(232102210141)

Strawberry leaf disease detection algorithm based on dynamic adaptive feature network

YU Fengjun¹(), ZHOU Xiaoping²^,^*(), GONG Tanfei³

¹ Department of Information Technology, Henan Judicial Police Vocational College, Zhengzhou 450046, China
² School of Electrical and Information Engineering, Zhengzhou University, Zhengzhou 450001, China
³ College of Information and Management Science, Henan Agricultural University, Zhengzhou 450046, China

Received:2025-04-01 Published:2026-08-25 Online:2026-09-14

摘要/Abstract

摘要：

为解决自然环境中草莓叶片病害检测场景复杂、小目标病害检测难度高的问题,提出一种基于动态自适应特征网络(DAFNet)的草莓叶片病害检测算法。首先,在骨干网络中构建多维特征注意力(MFA)机制,以增强对病害特征的关注;然后,在颈部网络设计语义增强模块(SRM)和纹理增强模块(TRM),以提升模型对复杂背景下细节信息的提取能力。此外,采用DySample上采样模块对不同层次的特征图进行自适应优化,以提高对小尺度病害目标的识别精度。结果表明,DAFNet在草莓叶片病害数据集上的平均精度均值(mAP)达到91.6%,相比基线模型提升2.0百分点,同时保持较好的实时检测性能,为实际应用中的草莓叶片病害检测任务提供了有效的解决方案。

关键词: 草莓叶片病害, 目标检测, 注意力机制, DySample上采样模块

Abstract:

Aiming at the problems of complex detection scenarios and difficult detection of small-target diseases in strawberry leaf disease detection under natural environments, we propose a strawberry leaf disease detection algorithm based on a dynamic adaptive feature network (DAFNet) in this paper. Firstly, a multi-dimensional feature attention (MFA) mechanism is constructed in the backbone network to strengthen the model’s attention to disease features. Secondly, a semantic refinement module (SRM) and a texture refinement module (TRM) are designed in the neck network to enhance the model’s capability of extracting detailed information under complex backgrounds. Furthermore, the DySample upsampling module is adopted to adaptively optimize multi-level feature maps, so as to improve the recognition accuracy of small-scale disease targets. Experimental results demonstrate that the proposed DAFNet achieves a mean average precision (mAP) of 91.6% on the strawberry leaf disease dataset, which is 2.0 percentage points higher than the baseline model. Meanwhile, it maintains favorable real-time detection performance, providing an effective solution for practical strawberry leaf disease detection tasks.

Key words: strawberry leaf disease, target detection, attention mechanism, DySample upsampling module

中图分类号:

余风军, 周晓平, 龚谭飞. 基于动态自适应特征网络的草莓叶片病害检测算法[J]. 浙江农业学报, 2026, 38(8): 1660-1673.

YU Fengjun, ZHOU Xiaoping, GONG Tanfei. Strawberry leaf disease detection algorithm based on dynamic adaptive feature network[J]. Acta Agriculturae Zhejiangensis, 2026, 38(8): 1660-1673.

图/表 13

图1 草莓叶片病害图像示例

Fig.1 Example images of strawberry leaf diseases

表1 各类样本的数量分布

Table 1 Quantity distribution of samples

类别 Category	数量分布Quantity distribution
类别 Category	训练集 Training set	验证集 Validation set	测试集 Test set
角叶斑病Angular leaf spot	13 657	3 965	1 951
钙素缺乏症Calcium deficiency	3 125	890	446
叶斑病Leaf spot	12 254	3 546	1 756
白粉病Powdery mildew	9 415	2 687	1 339
总计Total	38 451	11 088	5 492

图2 DAFNet的网络结构图 MFA,多维特征注意力;SRM,语义增强模块;TRM,纹理增强模块。

Fig.2 Network structure diagram of DAFNet MFA,Multi-dimensioal feature attention; SRM,Semantic refinement module; TRM,Texture refinement module.

图3 三元注意力网络结构图

Fig.3 Triplet Attention network structure diagram

图4 多维特征注意力网络结构图

Fig. 4 Network structure diagram of multidimensional feature attention

图5 语义增强模块和纹理增强模块的网络结构图

Fig. 5 Network structure diagram of semantic refinement module and texture refinement module

图6 DySample上采样流程

Fig.6 DySample upsampling process

表2 不同注意力机制的性能对比

Table 2 Comparison of performance of different attention mechanisms

注意力机制 Attention mechanism	P/%	R/%	mAP/%	IS/s^-1
无Additive free	90.7	89.3	90.4	60.3
CBAM	90.1	86.8	89.3	58.4
BAM	89.8	86.5	89.1	58.7
EMA	91.8	88.7	90.7	58.3
MFA	92.7	90.9	91.6	58.2

表3 消融试验结果

Table 3 Results of ablation experiments

MFA	SRM+TRM	DySample	P/%	R/%	mAP/%	IS/s^-1	Params/10⁶	FLOPs/10⁹
×	×	×	89.8	88.9	89.6	60.3	31.8	79.0
√	×	×	91.2	89.8	90.9	58.5	32.2	80.3
×	√	×	90.9	89.5	90.4	58.8	32.0	80.1
×	×	√	90.4	89.2	89.9	59.3	31.9	79.8
√	√	×	91.8	90.3	91.1	57.9	32.4	81.2
×	√	√	91.3	90.1	91.2	58.0	32.2	80.9
√	×	√	91.5	90.5	91.4	58.2	32.3	81.2
√	√	√	92.7	90.9	91.6	57.3	32.5	81.5

图7 不同模型的P(准确率)-R(召回率)曲线对比

Fig.7 Comparison of P(precision)-R(recall) curves between models

表4 对比实验结果

Table 4 Results of comparative experiments

网络模型Network model	P	R	mAP/%	IS/s^-1	Params/10⁶	FLOPs/10⁹
Faster R-CNN	86.4	85.1	85.7	6.8	102.7	168.3
SSD	85.6	83.8	84.9	17.6	29.4	99.1
FCOS	89.8	88.9	89.6	60.3	31.8	79.0
YOLOv9-M	88.7	86.8	87.4	71.3	20.0	76.3
YOLOv10-M	90.6	87.2	89.1	72.1	15.4	59.1
YOLOv11-M	90.9	88.1	90.3	70.9	20.1	68.0
KTD-YOLOv8	89.2	86.3	87.5	82.6	13.4	—
MCDCNet	88.5	87.0	88.2	45.2	44.85	184.9
LGM-Net	91.3	90.2	90.8	15.4	82.1	—
YOLOv8n-vegetable	88.2	86.1	88.7	119.1	9.1	—
DAFNet	92.7	90.9	91.6	57.3	32.5	81.5

图8 不同算法检测结果的可视化对比

Fig. 8 Visual comparison of detection results of different algorithms

图9 不同场景下草莓叶片病害检测效果对比

Fig. 9 Comparison of detection effect of strawberry leaf disease in different scenarios

参考文献 28

[1]	刘继展, 吴硕. 草莓全程生产机械化技术与装备研究进展[J]. 农业机械学报, 2021, 52(5): 1-16.
	Liu J Z, Wu S. Research progress and prospect of strawberry whole-process farming mechanization technology and equipment[J]. Transactions of the Chinese Society for Agricultural Machinery, 2021, 52(5): 1-16.
[2]	叶琪, 王丽芬, 马明涛, 等. 基于改进YOLO v8的草莓病害检测方法[J]. 江苏农业科学, 2024, 52(20): 250-259.
	Ye Q, Wang L F, Ma M T, et al. Strawberry disease detection method based on improved YOLO v8[J]. Jiangsu Agricultural Sciences, 2024, 52(20): 250-259.
[3]	Ahmad I, Yang Y Y, Yue Y, et al. Deep learning based detector YOLOv5 for identifying insect pests[J]. Applied Sciences, 2022, 12(19): 10167.
[4]	Pang H T, Zhang Y T, Cai W M, et al. A real-time object detection model for orchard pests based on improved YOLOv4 algorithm[J]. Scientific Reports, 2022, 12: 13557.
[5]	程嘉瑜, 陈妙金, 李彤, 等. 基于改进Faster-RCNN网络的无人机遥感影像桃树检测[J]. 浙江农业学报, 2024, 36(8): 1909-1919.
	Cheng J Y, Chen M J, Li T, et al. Detection of peach trees in unmanned aerial vehicle (UAV) images based on improved Faster-RCNN network[J]. Acta Agriculturae Zhejiangensis, 2024, 36(8): 1909-1919
[6]	刘君, 王学伟. 基于YOLO的番茄病虫害识别算法[J]. 中国瓜菜, 2020, 33(9): 18-22, 38.
	Liu J, Wang X W. Tomato disease and pest detection algorithm based on YOLO convolutional neural network[J]. China Cucurbits and Vegetables, 2020, 33(9): 18-22, 38.
[7]	杨堃, 范习健, 薄维昊, 等. 基于视觉加强注意力模型的植物病虫害检测[J]. 南京林业大学学报(自然科学版), 2023, 47(3): 11-18.
	Yang K, Fan X J, Bo W H, et al. Plant disease and pest detection based on visiual attention enhancement[J]. Journal of Nanjing Forestry University(Natural Sciences Edition), 2023, 47(3): 11-18.
[8]	李就好, 林乐坚, 田凯, 等. 改进Faster R-CNN的田间苦瓜叶部病害检测[J]. 农业工程学报, 2020, 36(12): 179-185.
	Li J H, Lin L J, Tian K, et al. Detection of leaf diseases of balsam pear in the field based on improved Faster R-CNN[J]. Transactions of the Chinese Society of Agricultural Engineering, 2020, 36(12): 179-185.
[9]	Qin K B, Zhang J, Hu Y. Identification of insect pests on soybean leaves based on SP-YOLO[J]. Agronomy, 2024, 14(7): 1586.
[10]	Li Y, Wang J C, Wu H R, et al. Detection of powdery mildew on strawberry leaves based on DAC-YOLOv4 model[J]. Computers and Electronics in Agriculture, 2022, 202: 107418.
[11]	Xue Z Y, Xu R J, Bai D, et al. YOLO-tea: a tea disease detection model improved by YOLOv5[J]. Forests, 2023, 14(2): 415.
[12]	Xie X Y, Ma Y, Liu B, et al. A deep-learning-based real-time detector for grape leaf diseases using improved convolutional neural networks[J]. Frontiers in Plant Science, 2020, 11: 751.
[13]	郭阳, 许贝贝, 陈桂鹏, 等. 基于卷积神经网络的水稻虫害识别方法[J]. 中国农业科技导报, 2021, 23(11): 99-109.
	Guo Y, Xu B B, Chen G P, et al. Rice insect pest recognition method based on convolutional neural network[J]. Journal of Agricultural Science and Technology, 2021, 23(11): 99-109.
[14]	刘鹏, 张天翼, 冉鑫, 等. 基于PBM-YOLOv8的水稻病虫害检测[J]. 农业工程学报, 2024, 40(20): 147-156.
	Liu P, Zhang T Y, Ran X, et al. Detecting rice disease using PBM-YOLOv8[J]. Transactions of the Chinese Society of Agricultural Engineering, 2024, 40(20): 147-156.
[15]	王慧云, 赵俊生, 王禹, 等. 面向无人边防的复杂环境遮挡小目标检测算法[J]. 电子测量技术, 2024, 47(21): 168-177.
	Wang H Y, Zhao J S, Wang Y, et al. Complex environment occlusion small target detection algorithm for unmanned border defense[J]. Electronic Measurement Technology, 2024, 47(21): 168-177.
[16]	张中华, 贾伟宽, 邵文静, 等. 优化FCOS网络复杂果园环境下绿色苹果检测模型[J]. 光谱学与光谱分析, 2022, 42(2): 647-653.
	Zhang Z H, Jia W K, Shao W J, et al. Green apple detection based on optimized FCOS in orchards[J]. Spectroscopy and Spectral Analysis, 2022, 42(2): 647-653.
[17]	徐立鸿, 黄志尊, 龙伟, 等. 基于MobileViT-CBAM-BiLSTM的开放式养殖环境鱼群摄食强度分类模型[J]. 农业机械学报, 2024, 55(11): 147-153.
	Xu L H, Huang Z Z, Long W, et al. Classification model of fish feeding intensity based on MobileViT-CBAM-BiLSTM[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(11): 147-153.
[18]	张寅升, 宋曾林, 王海燕. 基于注意力机制的水果新鲜度检测可解释模型[J]. 中国食品学报, 2024, 24(10): 28-36.
	Zhang Y S, Song Z L, Wang H Y. Fruit freshness detection explainable model based on attention mechanism[J]. Journal of Chinese Institute of Food Science and Technology, 2024, 24(10): 28-36.
[19]	Misra D, Nalamada T, Arasanipalai A U, et al. Rotate to attend: convolutional triplet attention module[C]//2021 IEEE Winter Conference on Applications of Computer Vision (WACV). January 3-8, 2021, Waikoloa, HI, USA. New York: IEEE, 2021: 3138-3147.
[20]	罗斌, 李家超, 周亚男, 等. 基于改进YOLO v8的轻量化稻瘟病孢子检测方法[J]. 农业机械学报, 2024, 55(11): 32-38.
	Luo B, Li J C, Zhou Y N, et al. Lightweight rice blast spores detection method based on improved YOLO v8[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(11): 32-38.
[21]	张岚, 邢博闻, 李彩, 等. 采用改进SSD网络的海参目标检测算法[J]. 农业工程学报, 2022, 38(8): 297-303.
	Zhang L, Xing B W, Li C, et al. Algorithm for detecting sea cucumbers based on improved SSD[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(8): 297-303.
[22]	Gui H T, Su T, Jiang X H, et al. FS-YOLOv9: a frequency and spatial feature-based YOLOv9 for real-time breast cancer detection[J]. Academic Radiology, 2025, 32(3): 1228-1240.
[23]	Qiu X Y, Chen Y J, Cai W H, et al. LD-YOLOv10: a lightweight target detection algorithm for drone scenarios based on YOLOv10[J]. Electronics, 2024, 13(16): 3269.
[24]	Sharma A, Kumar V, Longchamps L. Comparative performance of YOLOv8, YOLOv9, YOLOv10, YOLOv11 and Faster R-CNN models for detection of multiple weed species[J]. Smart Agricultural Technology, 2024, 9: 100648.
[25]	He Y L, Peng Y F, Wei C Y, et al. Automatic disease detection from strawberry leaf based on improved YOLOv8[J]. Plants, 2024, 13(18): 2556.
[26]	Liu B, Huang X L, Sun L M, et al. MCDCNet: Multi-scale constrained deformable convolution network for apple leaf disease detection[J]. Computers and Electronics in Agriculture, 2024, 222: 109028.
[27]	Qu Y M, Yu S B, Yang J. LGM-net: wheat pest and disease detection network based on local global information interaction and multi-level feature fusion[J]. IEEE Access, 2024, 12: 180267-180278.
[28]	Wang X W, Liu J. Vegetable disease detection using an improved YOLOv8 algorithm in the greenhouse plant environment[J]. Scientific Reports, 2024, 14: 4261.

基于动态自适应特征网络的草莓叶片病害检测算法

Strawberry leaf disease detection algorithm based on dynamic adaptive feature network

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 28

相关文章 12

编辑推荐

Metrics

本文评价

[1]	李功磊, 黄可玥, 刘英豪, 张陈文, 李梓桐, 伍德林. 基于YOLO-AIFI的蓝莓成熟度与果簇识别方法[J]. 浙江农业学报, 2026, 38(6): 1258-1270.
[2]	张郭, 周庆辉, 何胜喜. 基于SOLOv2改进的苹果采摘机器人实例分割算法[J]. 浙江农业学报, 2026, 38(3): 588-599.
[3]	汤奥冉, 金秀, 王坦, 饶元, 李佳佳, 张武. 基于弯曲大豆植株主茎骨架重构的生理株高测量方法[J]. 浙江农业学报, 2025, 37(2): 466-479.
[4]	李萌民, 刘朔, 欧阳宇, 张鹏. 基于YOLOv8n的高效轻量化柑橘叶片病害检测模型[J]. 浙江农业学报, 2025, 37(10): 2198-2208.
[5]	郭秀明, 王大伟, 刘升平, 诸叶平, 刘晓辉, 林克剑, 王佳宇, 李非. 基于深度学习的近地面草原鼠洞识别计数关键问题研究与应用[J]. 浙江农业学报, 2024, 36(9): 2146-2154.
[6]	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.
[7]	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.
[8]	李颀, 李煜哲. 弱光条件下猪舍清洗目标检测[J]. 浙江农业学报, 2023, 35(9): 2240-2249.
[9]	李荣鹏, 买买提·沙吾提, 盛艳芳, 何旭刚. 基于CA-MobileNet-V2的核桃病害识别与应用[J]. 浙江农业学报, 2023, 35(12): 2977-2987.
[10]	王英允, 龙燕, 杨智优, 黄铝文. 基于改进U-Net网络的苹果叶部病害语义分割方法[J]. 浙江农业学报, 2023, 35(11): 2731-2741.
[11]	周品志, 裴悦琨, 魏冉, 张永飞, 谷宇. 基于YOLOV4模型的果园樱桃实时检测研究[J]. 浙江农业学报, 2022, 34(11): 2522-2532.
[12]	张宁, 吴华瑞, 韩笑, 缪祎晟. 基于多尺度和注意力机制的番茄病害识别方法[J]. 浙江农业学报, 2021, 33(7): 1329-1338.