[1]刘希岩,赵秋实,杨锦锋,等.基于YOLO-GLBW模型的复杂田间环境下玉米田杂草检测[J].江苏农业学报,2026,42(08):1584-1596.[doi:doi:10.3969/j.issn.1000-4440.2026.08.008]
　LIU Xiyan,ZHAO Qiushi,YANG Jinfeng,et al.Weed detection in maize fields under complex field conditions based on the YOLO-GLBW model[J].,2026,42(08):1584-1596.[doi:doi:10.3969/j.issn.1000-4440.2026.08.008]
点击复制

基于YOLO-GLBW模型的复杂田间环境下玉米田杂草检测()

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
42
期数:
2026年08期
页码:
1584-1596
栏目:
农业信息工程
出版日期:
2026-08-31

文章信息/Info

Title:
Weed detection in maize fields under complex field conditions based on the YOLO-GLBW model
作者:
刘希岩1; 赵秋实1; 杨锦锋1; 刘铱凡11; 隋新2
(1.哈尔滨理工大学计算机科学与技术学院,黑龙江哈尔滨150080；2.黑龙江省农垦科学院农业工程研究所,黑龙江哈尔滨150086)
Author(s):
LIU Xiyan1; ZHAO Qiushi1; YANG Jinfeng1; LIU Yifan1; SUI Xin2
(1.College of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150080, China;2.Institute of Agricultural Engineering, Heilongjiang Academy of Land Reclamation Sciences, Harbin 150086, China)
关键词:
YOLOv8n模型;  玉米田; 杂草; 目标检测
Keywords:
YOLOv8n model; maize field; weeds; object detection
分类号:
TP391.41；S513
DOI:
doi:10.3969/j.issn.1000-4440.2026.08.008
文献标志码:
A
摘要:
针对实际农田复杂环境中玉米苗与杂草相互遮挡、形态相似、杂草分布密集,导致杂草目标定位困难,且现有检测模型存在长程依赖建模不足、检测精度不高、推理效率偏低等问题,本研究以YOLO v8n为基础,提出一种改进模型YOLO-GLBW。首先,在主干网络特征提取模块(C2f)以及颈部网络全部C2f模块中嵌入幽灵卷积(GhostConv)结构,有效降低模型的参数量与运算量,同时保留完备的特征表达能力。在快速空间金字塔池化(SPPF)模块中引入大型可分离核注意力机制(LSKA),有效弥补模型长程依赖不足的缺陷,增强对小目标的感知能力。其次,选用加权双向特征金字塔网络(BiFPN),提升模型对多尺度目标的感知能力与定位精度。最后,引入明智交并比(WIoU)损失函数,优化复杂场景下的定位精度与分类效果,加快模型的收敛速度,增强泛化能力。本次试验采用自行构建的田间玉米苗图像与杂草图像数据集开展测试,结果表明,YOLO-GLBW模型mAP50达83.9%,模型的参数量为2.47 M,1.0×109次浮点运算次数(GFLOPs)为6.3,与原始模型YOLOv8相比,YOLO-GLBW模型精确率、召回率、mAP50和mAP50-95分别提高了1.4个百分点、2.1个百分点、1.6个百分点、0.5个百分点,同时参数量和运算量分别降低17.9%和22.2%。YOLO-GLBW模型在显著降低运算开销、实现模型轻量化的同时,进一步提升了目标检测精度。面对田间玉米苗与杂草密集分布、形态相近、相互遮挡的复杂场景,该模型仍可保持稳定的检测精度,解决了模型识别精度与效率难以兼顾的问题,具备良好的田间落地应用价值。
Abstract:
To address the issues of mutual occlusion between maize seedlings and weeds, similar morphology, and dense weed distribution in complex field environments, which lead to difficulties in weed target localization, as well as the problems of insufficient long-range dependency modeling, low detection accuracy, and low inference efficiency in existing detection models, this study proposed an improved model, YOLO-GLBW, based on YOLOv8n. First, we embedded the ghost convolution (GhostConv) structure into the C2f modules of both the backbone feature extraction network and all C2f modules of the neck network, which effectively reduced the model’s parameter count and computational cost while preserving its full feature representation capability. We introduced the large separable kernel attention (LSKA) mechanism into the fast spatial pyramid pooling (SPPF) module, which effectively compensated for the model’s deficiency in long-range dependency and enhanced its perception capability for small targets. Second, we adopted a bidirectional feature pyramid network (BiFPN) to improve multi-scale target detection and localization performance. Finally, we introduced the wise intersection over union (WIoU) loss function to optimize localization and classification performance in complex scenarios, accelerate the convergence speed, and enhance the generalization ability. In this experiment, we tested the proposed model on a self-constructed dataset of maize seedling and weed images collected from field environments. The results showed that YOLO-GLBW achieved mAP50 of 83.9%, with 2.47 M parameters and a computational cost of 6.3 GFLOPs (1.0×109 floating-point operations). Compared with the original YOLOv8 model, YOLO-GLBW improved precision, recall, mAP50, and mAP50-95 by 1.4 percentage points, 2.1 percentage points, 1.6 percentage points, and 0.5 percentage points, respectively, while reducing the parameter count and computational cost by 17.9% and 22.2%, respectively. The YOLO-GLBW model significantly reduced computational cost and achieved a lightweight design, while further improving object detection accuracy. In the face of complex scenarios characterized by dense distribution, morphological similarity, and mutual occlusion between maize seedlings and weeds, the model maintains stable detection accuracy, resolves the trade-off between recognition precision and efficiency, and demonstrates considerable potential for practical field deployment.

参考文献/References:

[1]马延华,孙德全,李绥艳,等. 黑龙江省玉米地方品种主要农艺性状综合评价及优异种质资源筛选[J]. 作物杂志,2024(4):103-112.
[2]Hasan A S M M, Sohel F, Diepeveen D, et al. A survey of deep learning techniques for weed detection from images[J]. Computers and Electronics in Agriculture,2021,184:106067.
[3]肖立同,王熙. 基于改进YOLO算法的玉米苗间杂草检测模型研究[J]. 农机化研究,2025,47(8):10-16.
[4]付豪,赵学观,翟长远,等. 基于深度学习的杂草识别方法研究进展[J]. 中国农机化学报,2023,44(5):198-207.
[5]闫磊. 不同种植模式下玉米田杂草群落的演变及生态防控试验[J]. 种子科技,2025,43(16):5-7.
[6]唐伯青,赵大勇,熊锋,等. 基于视觉自注意力模型的苗期玉米与杂草检测方法[J]. 南京农业大学学报,2024,47(4):772-781.
[7]何海军,张珺栩. 基于改进Faster-RCNN的有害杂草识别技术分析[J]. 集成电路应用,2025,42(2):412-414.
[8]王靖,姜文刚,程耀,等. 基于Lab颜色空间改进U-Net的稻田杂草分割方法[J]. 中国农机化学报,2025,46(5):148-154.
[9]章磊,冷欣,陈佳凯,等. 基于改进RT-DETR模型的油菜田间杂草识别研究[J]. 湖北农业科学,2025,64(8):1-9.
[10]张天宇,韩静,廖洪晖,等. 基于优化YOLO v5算法的玉米苗间杂草检测研究[J]. 中国农机化学报,2025,46(7):220-225.
[11]孟志永,贾雅微,张秀清,等. 基于WAAP-YOLO的玉米伴生杂草检测模型[J]. 河北科技大学学报,2025,46(4):386-394.
[12]Li L, Sun R, Xu Y F. Design and optimization of a new corn-weed detection model with YOLO v8-GAS based on artificial intelligence[J]. Journal of Real-Time Image Processing,2025,22(5):167.
[13]Ma C R, Chi G, Ju X P, et al. YOLO-CWD:a novel model for crop and weed detection based on improved YOLO v8[J]. Crop Protection,2025,192:107169.
[14]文韬,王天一. 基于改进YOLO v8n的玉米地杂草检测[J]. 软件工程,2025,28(3):6-10.
[15]Su J, Song Z, Wang X J, et al. Maize seedling and weed detection using BFSL-YOLO v8[J]. World Scientific Research Journal,2025,11(3):20-28.
[16]Feng P B, Nie Z G, Li G. Field-deployable lightweight YOLO v8n for real-time detection and counting of maize seedlings using UAV RGB imagery[J]. Frontiers in Plant Science,2025,16:1639533.
[17]Gao J X, Tan F, Cui J P, et al. Unmanned aerial vehicle image detection of maize-YOLO v8n seedling leakage[J]. Frontiers in Plant Science,2025,16:1569229.
[18]樊湘鹏,张岩琪,周硕,等. 基于改进YOLO v8s与RGB-D信息融合的番茄采摘机器人识别与定位方法[J]. 农业工程学报,2025,41(15):106-116.
[19]李亚,蒋晨,王海瑞,等. 基于EDW-YOLO v8的棉花叶片病害检测[J]. 华中农业大学学报,2025,44(5):189-197.
[20]Han K, Wang Y H, Tian Q, et al. GhostNet:more features from cheap operations[C]//CVPR. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle,USA:IEEE,2020.
[21]Lau K W, Po L M, Rehman Y A U. Large separable kernel attention:rethinking the large kernel attention design in CNN[J]. Expert Systems with Applications,2024,236:121352.
[22]靳怡婷,李俊萩,张晴晖. 基于改进YOLO v7-tiny轻量化模型的复杂环境下柑橘果实识别方法[J]. 中国农机化学报,2025,46(9):153-162.
[23]肖紫娟. 基于改进YOLO v7的小目标检测算法研究[D]. 哈尔滨:黑龙江大学,2025.
[24]Tan M X, Pang R M, Le Q V. EfficientDet:scalable and efficient object detection[C]//CVPR. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle,USA:IEEE,2020.
[25]向琪琪,陈中举,李嘉诚,等. 基于改进YOLO v8s的自然环境下野生菌识别算法[J]. 中国农机化学报,2026,47(1):100-107.
[26]Tong Z J, Chen Y H, Xu Z W, et al. Wise-IoU:bounding box regression loss with dynamic focusing mechanism[EB/OL]. (2023-01-24)
[2025-08-08]. https://arxiv.org/abs/2301.10051.
[27]邹承明,薛榕刚. 融合GIoU和Focal loss的YOLO v3目标检测算法[J]. 计算机工程与应用,2020,56(24):214-222.
[28]Zheng Z H, Wang P, Liu W, et al. Distance-IoU loss:faster and better learning for bounding box regression[J]. Proceedings of the AAAI Conference on Artificial Intelligence,2020,34(7):12993-13000.
[29]Gevorgyan Z. SIoU loss:more powerful learning for bounding box regression[EB/OL]. (2022-05-25)
[2025-08-08]. https://arxiv.org/abs/2205.12740.
[30]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.
[31]李丽,张官明,张云峰,等. 基于YOLO v5-MNv4S与RANSAC-GN的柑橘姿态估计方法[J]. 农业机械学报,2025,56(10):558-566.
[32]丁志康,兰玉彬,韩鑫,等. 基于改进YOLO v7-tiny的农药残留消解过程多品种果蔬精准识别方法[J]. 农业机械学报,2025,56(10):725-735.
[33]韩科立,王振坤,余永峰,等. 基于改进YOLO 11n模型的棉花田间复杂环境障碍物检测方法[J]. 农业机械学报,2025,56(5):111-120.

相似文献/References:

[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.[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,(08):323.[doi:doi:10.3969/j.issn.1000-4440.2025.02.013]

备注/Memo

备注/Memo:
收稿日期：2026-01-05基金项目：黑龙江省自然科学基金项目(LH2023F048)作者简介：刘希岩(2000-),男,黑龙江哈尔滨人,硕士研究生,主要研究方向为计算机视觉智慧农业。(E-mail)1294585859@qq.com通讯作者：杨锦锋,(E-mail)yangjinfeng@hrbust.edu.cn
更新日期/Last Update: 2026-09-16