基于Swin-TDL算法的果园环境下葡萄病害检测方法

doi:10.13733/j.jcam.issn.2095-5553.2024.08.034

摘要/Abstract

摘要： 为在果园复杂环境因素下准确检测葡萄病害，提出一种基于Swin Transformer的葡萄病害检测模型Swin-TDL。采用Kmeans++聚类算法计算模型输入图像的锚框以提高检测精度；以Swin Transformer网络作为Swin-TDL的骨干网络更准确地获取目标特征信息；特征金字塔网络和路径聚合网络用于融合骨干网络中不同深度的特征层信息以增强检测目标的语义信息和定位信息；使用SIoU损失函数作为边界回归预测损失函数用于提高训练的速度和模型推理的准确性；使用Soft-NMS对目标边界框后处理以提高遮挡及重叠目标的检出率。在田间葡萄病害数据集中进行模型训练和性能测试，结果表明，Swin-TDL模型的平均精度均值为92.7%，平均检测时间为15.3 ms，综合性能优于对比检测算法，可以为葡萄植保装备研究提供参考。

关键词: 葡萄病害检测, 果园复杂环境, 图像增强, 深度学习, Swin Transformer

Abstract: In order to accurately detect grape diseases under complex environmental factors in orchards, a grape disease detection model Swin-TDL based on Swin Transformer is proposed. Kmeans++ clustering algorithm is used to calculate the anchor frames of the model input images to improve the detection accuracy. The Swin Transformer network is used as the backbone network of Swin-TDL for more accurate acquisition of target feature information. Feature pyramid networks and path aggregation networks are used to fuse information from feature layers of different depths in the backbone network to enhance the semantic and localization information of detection targets. The SIoU loss function is used as a boundary regression prediction loss function for improving the speed of training and the accuracy of model inference. Soft-NMS is used to post‑process the target bounding boxes to improve the detection rate of occluded and overlapped targets. Finally, model training and performance testing were carried out in the field grape disease dataset. The experimental results showed that the average accuracy of the Swin-TDL model was 92.7% and the average detection time was 15.3 ms. The comprehensive performance of Swin-TDL was better than the comparative detection algorithm, which could provide a reference for the research of grape plant protection equipment.

Key words: grape disease detection, orchard complex environment, image enhancement, deep learning, Swin Transformer

中图分类号:

胡国玉, 刘广, 周星光, 董娅兰, 周建平, . 基于Swin-TDL算法的果园环境下葡萄病害检测方法[J]. 中国农机化学报, 2024, 45(8): 234-239.

Hu Guoyu, , Liu Guang, Zhou Xingguang, Dong Yalan, Zhou Jianping, . Detection method of grape disease in orchard environment based on Swin-TDL algorithm [J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 234-239.

参考文献

[ 1 ] Jaisakthi S M， Mirunalini P， Thenmozhi D. Grape leaf disease identification using machine learning techniques [C]. 2019 International Conference on Computational Intelligence in Data Science （ICCIDS）. IEEE， 2019: 1-6.
[ 2 ] Ge L， Zou K， Zhou H， et al. Three dimensional apple tree organs classification and yield estimation algorithm based on multi‑features fusion and support vector machine [J]. Information Processing in Agriculture， 2022， 9（3）: 431-442.
[ 3 ] Zhu J， Wu A， Wang X， et al. Identification of grape diseases using image analysis and BP neural networks [J]. Multimedia Tools and Applications， 2020， 79（21）: 14539-14551.
[ 4 ] Tian Y， Yang G， Wang Z， et al. Apple detection during different growth stages in orchards using the improved YOLO-V3 model [J]. Computers and Electronics in Agriculture， 2019， 157: 417-426.
[ 5 ] Kerkech M， Hafiane A， Canals R. Deep leaning approach with colorimetric spaces and vegetation indices for vine diseases detection in UAV images [J]. Computers and Electronics in Agriculture， 2018， 155: 237-243.
[ 6 ] 何欣，李书琴，刘斌. 基于多尺度残差神经网络的葡萄叶片病害识别[J]. 计算机工程， 2021， 47（5）: 285-291， 300.
He Xin， Li Shuqin， Liu Bin. Identification of grape leaf diseases based on multi‑scale residual neural network [J]. Computer Engineering， 2021， 47（5）: 285-291， 300.
[ 7 ] Li H， Li C， Li G， et al. A real‑time table grape detection method based on improved YOLOv4-tiny network in complex background [J]. Biosystems Engineering， 2021， 212: 347-359.
[ 8 ] Han K， Wang Y， Chen H， et al. A survey on vision transformer [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2022， 45（1）: 87-110.
[ 9 ] Liu Z， Lin Y， Cao Y， et al. Swin transformer: Hierarchical vision transformer using shifted windows [C]. Proceedings of the IEEE/CVF International Conference on Computer Vision， 2021: 10012-10022.
[10] Li X， Li S. Transformer help CNN see better: A lightweight hybrid apple disease identification model based on transformers [J]. Agriculture， 2022， 12（6）: 884.
[11] Wang J， Zhang Z， Luo L， et al. SwinGD: A robust grape bunch detection model based on Swin transformer in complex vineyard environment [J]. Horticulturae， 2021， 7（11）: 492.
[12] Zaidi S S A， Ansari M S， Aslam A， et al. A survey of modern deep learning based object detection models [J]. Digital Signal Processing， 2022， 126: 103514.
[13] Zhao Z Q， Zheng P， Xu S， et al. Object detection with deep learning: A review [J]. IEEE Transactions on Neural Networks and Learning Systems， 2019， 30（11）: 3212-3232.
[14] Xue J， Cheng F， Li Y， et al. Detection of farmland obstacles based on an improved YOLOv5s algorithm by using CIoU and anchor box scale clustering [J]. Sensors， 2022， 22（5）: 1790.
[15] Ren Z， Lam E Y， Zhao J. Real‑time target detection in visual sensing environments using deep transfer learning and improved anchor box generation [J]. IEEE Access， 2020， 8: 193512-193522.
[16] Fu L， Duan J， Zou X， et al. Fast and accurate detection of banana fruits in complex background orchards [J]. IEEE Access， 2020， 8: 196835-196846.
[17] Li H， Li C， Li G， et al. A real‑time table grape detection method based on improved YOLOv4-tiny network in complex background [J]. Biosystems Engineering， 2021， 212: 347-359.
[18] Gevorgyan Z. SIoU loss: More powerful learning for bounding box regression [J]. arxiv preprint arxiv: 2205.12740， 2022.

[1]	刘宇平, 刘程飞, 赵平伟. 基于DeepLabv3—Faster R—CNN的水稻叶片病害检测方法#br#[J]. 中国农机化学报, 2025, 46(4): 108-113.
[2]	丁展博, 徐健, 朱耀麟, 张勇进, 刘晨雨. 基于轻量化YOLO模型的花椒簇检测研究[J]. 中国农机化学报, 2025, 46(4): 120-125.
[3]	陈浩宇, 苗玉彬. 基于GASF变换和深度学习的柑橘内部品质分析[J]. 中国农机化学报, 2025, 46(4): 133-138.
[4]	赵玉清, 焦雨杰, 李宏, 王天允, 李嘉舜, 张悦, . 改进ShuffleNet V1算法的咖啡豆分级方法研究[J]. 中国农机化学报, 2025, 46(4): 194-203.
[5]	吕宗旺, 王甜甜, 孙福艳, 祝玉华, . 基于改进YOLOv8m的小麦仓储粮虫检测方法[J]. 中国农机化学报, 2025, 46(3): 108-114.
[6]	吴坚, , 叶梦焱, 张同锋. 茶鲜叶智能分级装置设计与试验[J]. 中国农机化学报, 2025, 46(3): 139-145.
[7]	任晶晶, 张小勇, 贾伟宽. 基于多粒度特征感知的FoveaBox绿色苹果抗遮挡检测模型[J]. 中国农机化学报, 2025, 46(3): 182-187.
[8]	郭敬涛, 吕凤, 章慧婷, 杨彪, 刘大洋. 基于迁移学习与轻量化YOLOv5s的草莓目标检测方法[J]. 中国农机化学报, 2025, 46(3): 253-260.
[9]	许毓超, , 吴茜, , 张兵园, , , 周玲莉, , 任妮, , , 张美娜, , . 轻量级深度学习网络在农作物目标检测的应用进展[J]. 中国农机化学报, 2025, 46(3): 261-270.
[10]	吴秋兰, 陈雪飞, 陈超, 张峰, 王姝妹, 赵恒. 基于改进YOLOv5s的香菇菌棒污染识别方法[J]. 中国农机化学报, 2025, 46(2): 217-223.
[11]	戴敏, 孙文靖, 缪宏. 基于轻量化CBAM—GoogLeNet的辣椒病虫害识别[J]. 中国农机化学报, 2025, 46(2): 224-229.
[12]	彭雨侬, 柳平增, 张艳, . 基于深度学习的玉米生产过程知识图谱构建[J]. 中国农机化学报, 2025, 46(2): 245-252.
[13]	夏子林, 张新洲, 王文波, 夏先飞, 陈兰, 顾寄南. 基于机器视觉的蚕豆荚高精度检测方法研究[J]. 中国农机化学报, 2025, 46(1): 157-163.
[14]	李春雨, 郭肖琴, 杨晶晶, 李忠华. 基于深度学习的水果图像识别[J]. 中国农机化学报, 2025, 46(1): 198-203.
[15]	牛潘婷, 张宝林, 潘丽杰, 郭建鹏(. 基于改进GoogLeNet的玉米叶片病害识别及其可解释性研究[J]. 中国农机化学报, 2025, 46(1): 204-212.