[1]唐希延,郭文芳,杨晶晶,等.基于YOLO v8n改进的轻量化樱桃叶片病害检测模型[J].江苏农业学报,2026,42(07):1419-1428.[doi:doi:10.3969/j.issn.1000-4440.2026.07.013]
　TANG Xiyan,GUO Wenfang,YANG Jingjing,et al.A lightweight cherry leaf disease detection model improved based on YOLO v8n[J].,2026,42(07):1419-1428.[doi:doi:10.3969/j.issn.1000-4440.2026.07.013]
点击复制

基于YOLO v8n改进的轻量化樱桃叶片病害检测模型()

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
42
期数:
2026年07期
页码:
1419-1428
栏目:
农业信息工程
出版日期:
2026-07-31

文章信息/Info

Title:
A lightweight cherry leaf disease detection model improved based on YOLO v8n
作者:
唐希延; 郭文芳; 杨晶晶; 赵宇轩; 冯贵良
(河北北方学院信息科学与工程学院,河北张家口075000)
Author(s):
TANG Xiyan; GUO Wenfang; YANG Jingjing; ZHAO Yuxuan; FENG Guiliang
(College of Information Science and Engineering, Hebei North University, Zhangjiakou 075000, China)
关键词:
樱桃; 病害检测; YOLO v8n模型; 轻量化模型
Keywords:
cherry; disease detection; YOLO v8n model; lightweight model
分类号:
TP391.4
DOI:
doi:10.3969/j.issn.1000-4440.2026.07.013
文献标志码:
A
摘要:
针对目前已有模型在樱桃病害检测中存在的识别精度低、计算复杂度高及移动端难部署等问题,本研究基于YOLO v8n模型提出一种轻量化樱桃叶片病害检测模型YOLO v8n-BFGC。首先,将YOLO v8n模型颈部网络中通道拼接（Concat）结构替换为双向特征金字塔网络(BiFPN)结构,增强多尺度特征融合能力,提升检测精度；其次,在骨干网络引入幽灵卷积(GhostConv)模块替换传统卷积模块,并采用C3Ghost模块,在保持精度的同时显著降低参数量和计算量。最后,在主干网络末端嵌入卷积块注意力(CBAM)模块,强化对关键特征的提取能力。试验结果表明,改进模型YOLO v8n-BFGC精确率(P)、召回率(R)和平均精度均值(mAP50)分别达到93.1%、89.7%和93.8%,相较于原始模型YOLO v8n,参数量减少63.3%、计算量降低49.4%。与YOLO主流模型相比,改进模型在精度与轻量化方面均具有优越性。该研究结果可为大棚环境下樱桃叶片病害的高效智能识别提供技术支撑。
Abstract:
In response to the problems of low recognition accuracy, high computational complexity, and difficult deployment on mobile terminals that existed in current models for cherry disease detection, this study proposed a lightweight cherry leaf disease detection model named YOLOv8n-BFGC based on YOLO v8n. Firstly, the Concat structure in the neck network of YOLO v8n model was replaced with a bidirectional feature pyramid network (BiFPN) structure to enhance the multi-scale feature fusion ability and improve the detection accuracy. Secondly, GhostConv module was introduced in the main network to replace the traditional convolution module, and the C3Ghost module was adopted, significantly reducing the parameter quantity and computational cost without sacrificing accuracy. Finally, the CBAM module was embedded at the end of the main network to strengthen the extraction ability of key features. Experimental results showed that the YOLO v8n-BFGC model achieved a precision (P) of 93.1%, a recall rate (R) of 89.7%, and a mean average precision (mAP50) of 93.8%. Compared with the original YOLO v8n model, the parameter quantity was reduced by 63.3%, and the computational cost was decreased by 49.4%. Compared with mainstream YOLO models, the improved model showed superiority in both accuracy and lightweightness. The results of this study can provide effective technical support for the efficient and intelligent identification of cherry leaf diseases in greenhouse environments.

参考文献/References:

[1]沈卓群. 樱桃种植机械化发展问题与对策研究[J]. 农业科技与装备,2025(2):73-74.
[2]马文江,李殿运. 费县大樱桃产业现状与发展建议[J]. 果农之友,2024(12):125-127.
[3]何明莉,张琪静,艾佳音,等. 鞍山市甜樱桃产业发展现状、问题及建议[J]. 中国果业信息,2024,41(11):35-38.
[4]钟银芹. 破局与重塑:山东烟台大樱桃物流体系的创新变革[J]. 食品界,2025(6):128-130.
[5]徐湾湾. 中国樱桃主要病虫害发生与绿色防控技术[J]. 中国果业信息,2025,42(6):70-71,74.
[6]宋长年,李红月,董瑞萍,等. 大樱桃病害防治技术研究进展[J]. 现代农业科技,2022(15):115-118,122.
[7]王迪聪,白晨帅,邬开俊. 基于深度学习的视频目标检测综述[J]. 计算机科学与探索,2021,15(9):1563-1577.
[8]Nandhini P, Mahaveerakannan R. An efficient treatment for cherry tree disease using the random forest classifier comparison with the decision tree algorithm[M]//Hung B T, Sekar M, Esi A, et al. Applications of Mathematics in Science and Technology. CRC Press,2025:636-640.
[9]蒋雪松,计恺豪,姜洪喆,等. 深度学习在林果品质无损检测中的研究进展[J]. 农业工程学报,2024,40(17):1-16.
[10]于涵,刘砚菊,冯迎宾. 基于YOLO v5s的轻量化车辆目标检测算法[J]. 工业控制计算机,2025,38(7):105-106,109.
[11]Dnmez E, nal Y, Kayhan H. Bacterial disease detection of cherry plant using deep features[J]. Sakarya University Journal of Computer and Information Sciences,2024,7(1):1-10.
[12]王宁,智敏. 深度学习下的单阶段通用目标检测算法研究综述[J]. 计算机科学与探索,2025,19(5):1115-1140.
[13]Redmon J, Divvala S, Girshick R, et al. You only look once:unified,real-time object detection[C]//IEEE. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas,NV,USA:IEEE,2016.
[14]Lin T Y, Goyal P, Girshick R, et al. Focal loss for dense object detection[C]//IEEE. 2017 IEEE International Conference on Computer Vision (ICCV). Venice,Italy:IEEE,2017.
[15]董元和,李恩泽,贾炎,等. YOLO-EZ:一种高效且轻量化的葡萄病害检测模型[J]. 湖北师范大学学报(自然科学版),2024,44(4):12-20.
[16]赵鑫,马明涛,王丽芬,等. 基于YOLO v5的轻量化苹果叶片病害检测[J]. 自动化应用,2024,65(18):146-148.
[17]王帅,王利众,朱丽平,等. 基于改进YOLO v5s的苹果病害检测技术研究[J]. 山西农业大学学报(自然科学版),2024,44(4):118-129.
[18]Bui T D, Do Le T M. Ghost-attention-YOLO v8:enhancing rice leaf disease detection with lightweight feature extraction and advanced attention mechanisms[J]. AgriEngineering,2025,7(4):93.
[19]Feng H R, Chen X Q, Duan Z Y. LCDDN-YOLO:lightweight cotton disease detection in natural environment,based on improved YOLO v8[J]. Agriculture,2025,15(4):421.
[20]Wang J F, Ma S Y, Wang Z T, et al. Improved lightweight YOLO v8 model for rice disease detection in multi-scale scenarios[J]. Agronomy,2025,15(2):445.
[21]Esmail M A, Wang J L, Wang Y H, et al. Resource-aware strategies for real-time multi-person pose estimation[J]. Image and Vision Computing,2025,155:105441.
[22]Khanam R, Hussain M. What is YOLO v5:a deep look into the internal features of the popular object detector[EB/OL].(2024-07-30)
[2025-08-20]. https://arxiv.org/abs/2407.20892.
[23]Han K, Wang Y H, Tian Q, et al. GhostNet:more features from cheap operations[C]//IEEE/CVF. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle,WA,USA:IEEE,2020.
[24]甄元元,刘平,金雨纯,等. 基于改进YOLO v8的葡萄叶片病害检测与识别[J]. 农业工程学报,2025,41(14):148-154.
[25]Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//IEEE. 2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus,OH,USA:IEEE,2014.
[26]梁绍鹏,李翰山. 基于SGE-YOLO v8的弹丸爆炸火光图像识别算法[J]. 探测与控制学报,2025,47(3):84-91,102.
[27]Yang L, Zhang R Y, Li L, et al. SimAM:a simple, parameter-free attention module for convolutional neu ral networks[C]//PMLR. International Conference on Machine Learning. Sanya:PMLR,2021.
[28]Li H, Liu J, Han K, et al. CAFM-enhanced YOLO v8:a two-stage optimization for precise strawberry disease detection in complex field conditions[J]. Applied Sciences,2025,15(18):10025.
[29]楚家,肖敏,周迅,等. 基于改进YOLO v8的复杂环境下农田害虫检测算法[J]. 江苏农业科学,2025,53(16):192-204.
[30]雷竣杰,周保平. 基于改进DCGAN的棉叶螨为害图像数据增强方法[J]. 江苏农业学报,2025,41(5):916-926.
[31]刘天真,苑迎春,滕桂法,等. 基于改进YOLO v4的自然场景下冬枣果实分类识别[J]. 江苏农业科学,2024,52(1):163-172.

相似文献/References:

[1]李仁杰,宋涛,高婕,等.基于改进YOLOv5的自然环境下番茄患病叶片检测模型[J].江苏农业学报,2024,(06):1028.[doi:doi:10.3969/j.issn.1000-4440.2024.06.009]
　LI Renjie,SONG Tao,GAO Jie,et al.Tomato diseased leaf detection model based on improved YOLOv5 in natural environment[J].,2024,(07):1028.[doi:doi:10.3969/j.issn.1000-4440.2024.06.009]
[2]杨如强,赵霞,张鑫.基于改进YOLOv11模型的柑橘叶片病害检测[J].江苏农业学报,2026,42(01):99.[doi:doi:10.3969/j.issn.1000-4440.2026.01.011]
　YANG Ruqiang,ZHAO Xia,ZHANG Xin.An improved YOLOv11 model for citrus leaf disease detection[J].,2026,42(07):99.[doi:doi:10.3969/j.issn.1000-4440.2026.01.011]

备注/Memo

备注/Memo:
收稿日期：2025-09-02基金项目：河北省高等学校科学研究项目(QN2024146)作者简介：唐希延(1999-),女,河南鹤壁人,硕士研究生,主要从事农业工程与信息技术研究。(E-mail)1657517446@qq.com通讯作者：冯贵良,(E-mail)6838710@qq.com
更新日期/Last Update: 2026-08-21