[1]李芳,危疆树,王玉超,等.基于改进YOLOv8n模型的辣椒病害检测方法[J].江苏农业学报,2025,(02):323-334.[doi:doi:10.3969/j.issn.1000-4440.2025.02.013]
　LI Fang,WEI Jiangshu,WANG Yuchao,et al.A chili disease detection method based on an improved YOLOv8n model[J].,2025,(02):323-334.[doi:doi:10.3969/j.issn.1000-4440.2025.02.013]
点击复制

基于改进YOLOv8n模型的辣椒病害检测方法()
分享到：

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2025年02期
页码:
323-334
栏目:
农业信息工程
出版日期:
2025-02-28

文章信息/Info

Title:
A chili disease detection method based on an improved YOLOv8n model
作者:
李芳1; 危疆树1; 王玉超2; 张尧1; 谢宇鑫1
(1.四川农业大学信息工程学院,四川雅安625014；2.四川农业大学机电学院,四川雅安625014)
Author(s):
LI Fang1; WEI Jiangshu1; WANG Yuchao2; ZHANG Yao1; XIE Yuxin1
(1.College of Information Engineering, Sichuan Agricultural University, Ya’an 625014, China;2.College of Mechanical and Electrical Engineering, Sichuan Agricultural University, Ya’an 625014, China)
关键词:
辣椒病害; YOLOv8n模型; 目标检测; Adown下采样模块; SlimNeck模块; Aux Head检测头
Keywords:
chili diseases; YOLOv8n model; target detection; Adown downsampling module; SlimNeck module; Aux Head
分类号:
S24；TP391.4
DOI:
doi:10.3969/j.issn.1000-4440.2025.02.013
文献标志码:
A
摘要:
为了解决辣椒病害检测速度慢,漏检率和误检率高的问题,本研究以YOLOv8n为基线模型,引入Adown下采样模块替代原模型骨干网络(Backbone)的卷积下采样层,引入SlimNeck模块将原模型颈部网络中的卷积层和特征聚合模块(C2f)替换为混合卷积模块(GSConv)和跨阶段部分网络(VoVGSCSP)模块,并利用辅助训练头Aux Head(Auxiliary head)融合原有的检测头,构建改进的YOLOv8n模型(YOLOv8n-ATA模型)。最后利用辣椒炭疽病、褐斑病、脐腐病和细菌性叶斑病等4种病害影像数据集对改进后的模型性能进行分析。结果表明，改进后模型的浮点计算量和模型大小比原YOLOv8n模型增加19.5%和10.2%,但模型对辣椒病害的识别精确率、平均精度均值mAP50和mAP50∶95分别提升2.6个百分点、2.9个百分点和2.9个百分点,同时每1 s传输帧数增加15.1%。因此,改进后的模型能够对辣椒病害进行有效识别,较好实现模型识别准确度与效率的平衡。
Abstract:
To solve the problems of slow detection speed, high missed detection rate, and high false detection rate in chili disease detection, this study used YOLOv8n as the baseline model and constructed an improved YOLOv8n model (YOLOv8n-ATA model). The Adown downsampling module was introduced to replace the convolutional downsampling layer of the Backbone of the original model. The SlimNeck module was introduced to replace the convolutional layer and feature aggregation module (C2f) in the neck network of the original model with the hybrid convolution module (GSConv) and the cross-stage partial network (VoVGSCSP) module. Moreover, the auxiliary training head Aux Head (Auxiliary head) was used to fuse the original detection head. Finally, the performance of the improved model was evaluated using the image datasets of four chili diseases, such as anthracnose, brown spot, blossom-end rot and bacterial leaf spot. The results showed that the floating-point calculations and size of the improved model were 19.5% and 10.2% higher than those of the original YOLOv8n model. However, the identification accuracy, mAP50 and mAP50∶95 of the model for chili diseases increased by 2.6 percentage points, 2.9 percentage points and 2.9 percentage points, respectively. At the same time, the number of frames per second increased by 15.1%. Therefore, the improved model can effectively identify chili diseases and better achieve the balance between accuracy and efficiency of model recognition.

参考文献/References:

[1]张子峰. 我国辣椒产业发展现状、主要挑战与应对之策[J]. 北方园艺,2023(14):153-158.
[2]乔立娟,赵帮宏,宗义湘,等. 我国辣椒产业发展现状、趋势及对策[J]. 中国蔬菜,2023(11):9-15.
[3]祝显萍. 辣椒种植技术及病害防治[J]. 种子科技,2023,41(21):93-95.
[4]REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once:unified,real-time object detection[C]//THE COMPUTER VISION FOUNDATION. 2016 IEEE conference on computer vision and pattern recognition (CVPR). Las Vegas,NV,USA:IEEE,2016:779-788.
[5]LIU W, ANGUELOV D, ERHAN D, et al. SSD:single shot MultiBox detector[M]//LEIBE B, MATAS J, SEBE N, et al. Computer vision-ECCV 2016 Part1. Cham:Springer International Publishing,2016:21-37.
[6]LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,42(2):318-327.
[7]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,2014:580-587.
[8]REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN:towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149.
[9]HE K M, ZHANG X Y, REN S Q, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(9):1904-1916.
[10]储鑫,李祥,罗斌,等. 基于改进YOLOv4算法的番茄叶部病害识别方法[J]. 江苏农业学报,2023,39(5):1199-1208.
[11]施杰,林双双,张威,等. 基于轻量化改进型YOLOv5s的玉米病虫害检测方法[J]. 江苏农业学报,2024,40(3):427-437.
[12]刘诗怡,胡滨,赵春. 基于改进YOLOv7的黄瓜叶片病虫害检测与识别[J]. 农业工程学报,2023,39(15):163-171.
[13]周维,牛永真,王亚炜,等. 基于改进的YOLOv4-GhostNet水稻病虫害识别方法[J]. 江苏农业学报,2022,38(3):685-695.
[14]邹玮,岳延滨,冯恩英,等. 基于YOLOv2的辣椒叶部蚜虫图像识别[J]. 山东农业大学学报(自然科学版),2023,54(5):700-709.
[15]尚俊平,张冬阳,席磊,等. 基于多尺度特征融合和注意力机制的辣椒病害识别模型[J]. 河南农业大学学报,2024,58(6):1021-1033.
[16]李西兴,陈佳豪,吴锐,等. 基于改进MaxViT的辣椒病害识别分类方法[J]. 华中农业大学学报,2024,43(2):123-133.
[17]姚飞. 基于卷积神经网络的辣椒病害检测方法研究与应用[D]. 重庆:重庆大学,2022.
[18]丁文宽. 基于卷积神经网络和机器视觉的辣椒检测与识别[D]. 天津:天津理工大学,2017.
[19]王启锟,王旭. 基于改进YOLOv5的辣椒目标检测算法研究[J]. 建模与仿真,2023,12(5):4654-4662.
[20]PRATAP V K, KUMAR N S. High-precision multiclass classification of chili leaf disease through customized EffecientNetB4 from chili leaf images[J]. Smart Agricultural Technology,2023,5:100295.
[21]KANAPARTHI K R, SUDHAKAR ILANGO S. A survey on training issues in chili leaf diseases identification using deep learning techniques[J]. Procedia Computer Science,2023,218:2123-2132.
[22]刘子洋,徐慧英,朱信忠,等. Bi-YOLO:一种基于YOLOv8n改进的轻量化目标检测算法[J]. 计算机工程与科学,2024,46(8):1444-1454.
[23]付辰伏,任力生,王芳. 基于改进YOLOv8的牛只行为识别与跟踪方法[J]. 农业机械学报,2024,55(5):290-301.
[24]WANG C Y, YEH I H, MARK LIAO H Y. YOLOv9:learning what you want to learn using programmable gradient information[Z/OL]. (2024-02-29). http://arxiv.org/pdf/2402.13616.
[25]LI H L, LI J, WEI H B, et al. Slim-neck by GSConv:a lightweight-design for real-time detector architectures[J]. Journal of Real-Time Image Processing,2024,21(3):62.
[26]WANG C Y, BOCHKOVSKIY A, MARK LIAO H Y. YOLOv7:trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//IEEE. 2023 IEEE/CVF conference on computer vision and pattern recognition (CVPR). Vancouver,BC,Canada:IEEE,2023:7464-7475.

备注/Memo

备注/Memo:
收稿日期：2024-06-18基金项目：四川省科技厅关键技术攻关项目(22ZDYF0095)作者简介：李芳(2001-),女,四川巴中人,硕士研究生,研究方向为机器视觉和目标检测。(E-mail)18116718920@163.com通讯作者：危疆树,(E-mail)weijiangshu66@163.com
更新日期/Last Update: 2025-03-27