嵌入式平台的番茄叶片病虫害检测模型

doi:10.3778/j.issn.1002-8331.2411-0193

摘要/Abstract

摘要： 针对真实田间环境下背景复杂、病虫害特征小以及现有检测模型难以部署的问题，提出一种RHEO-YOLO模型，并将改进模型部署于移动平台以实现实时检测。设计高效重参数聚合网络（RepELAN）模块以优化原有的C2f模块，在模型推理时将多分支网络结构重参数化，实现在不影响精度的同时降低模型参数量。使用高层筛选特征金字塔网络（high-level screening feature pyramid network，HSFPN）替换原有特征融合网络，增强模型对番茄叶片病虫害特征的表达能力。重新设计YOLOv8的检测头，并将损失函数更改为OIOU，加快模型收敛速度，并提高检测精度。实验结果表明改进后模型的精确率和召回率为89.4%和84.2%，平均检测精度为88.9%，同时参数量和模型大小分别降低约63%和60%，与其他目标检测模型相比，RHEO-YOLO模型在计算复杂度、检测精度方面具有更好的综合检测性能。在嵌入式设备Jetson Nano和Jetson Xavier NX上实时检测帧率分别为8.3?FPS和29.4?FPS，基本满足实时检测要求。改进后的模型能够有效检测自然环境下的番茄叶片病虫害目标，为番茄叶片病虫害检测算法改进和实际应用提供理论和技术支持。

关键词: 番茄叶片, YOLOv8, 轻量化, 目标检测, 嵌入式设备

Abstract: Aiming at the problems of complex background, small characteristics of pests and diseases and the difficulty in deploying existing detection models in the real field environment, a RHEO-YOLO model is proposed, and the deep learning model is deployed on a mobile platform to enable real-time detection. The re-parameterized efficient layer aggregation network (RepELAN) module is designed to optimize the original C2f module, and the multi-branch network structure is re-parameterized during model inference, so as to reduce the number of model parameters without affecting the accuracy. A high-level screening feature pyramid network (HSFPN) is employed to replace the original feature fusion network, enhancing the representational capability of the model for detecting tomato leaf pests and diseases. The detection head of YOLOv8 is redesigned, and the loss function is changed to OIOU to accelerate model convergence and improve detection accuracy. Experimental results show that the improved model achieves an accuracy of 89.4% and a recall rate of 84.2%, with an average detection accuracy of 88.9%. Furthermore, the number of parameters and model size are reduced by approximately 63% and 60%, respectively. Compared with other target detection models, the improved model demonstrates superior comprehensive detection performance in terms of computational complexity and detection accuracy. On the embedded devices Jetson Nano and Jetson Xavier NX, real-time detection frame rates of 8.3 FPS and 29.4 FPS are achieved, basically meeting the real-time detection requirements. The proposed model effectively detects tomato leaf pests and diseases in complex environments, providing theoretical and technical support for the enhancement and practical application of tomato leaf pest detection algorithms.

Key words: tomato leaf, YOLOv8, lightweight, target detection, embedded device

孙寿松, 李新凯, 张宏立, 何亮, 赵苓. 嵌入式平台的番茄叶片病虫害检测模型[J]. 计算机工程与应用, 2025, 61(16): 305-314.

SUN Shousong, LI Xinkai, ZHANG Hongli, HE Liang, ZHAO Ling. Embedded Platform for Tomato Leaf Pest Detection Model[J]. Computer Engineering and Applications, 2025, 61(16): 305-314.

参考文献

[1] TERENTEV A, DOLZHENKO V, FEDOTOV A, et al. Current state of hyperspectral remote sensing for early plant disease detection: a review[J]. Sensors, 2022, 22(3): 757.
[2] 黄英来, 姜忠良. 改进残差网络甜瓜叶片病害的识别研究[J]. 计算机工程与应用, 2024, 60(15): 189-197.
HUANG Y L, JIANG Z L. Research on identification of melon leaf diseases with improved residual network[J]. Computer Engineering and Applications, 2024, 60(15): 189-197.
[3] XU W T, ZHANG G X, DUAN Y. Farmland detection in synthetic aperture radar images with texture signature[J]. Journal of Applied Remote Sensing, 2014, 8(1): 084997.
[4] SINGH V, SHARMA N, SINGH S. A review of imaging techniques for plant disease detection[J]. Artificial Intelligence in Agriculture, 2020, 4: 229-242.
[5] MAES W H, STEPPE K. Perspectives for remote sensing with unmanned aerial vehicles in precision agriculture[J]. Trends in Plant Science, 2019, 24(2): 152-164.
[6] LI L Y, MU X H, MACFARLANE C, et al. A half-Gaussian fitting method for estimating fractional vegetation cover of corn crops using unmanned aerial vehicle images[J]. Agricultural and Forest Meteorology, 2018, 262: 379-390.
[7] ZHANG X, HAN L X, DONG Y Y, et al. A deep learning-based approach for automated yellow rust disease detection from high-resolution hyperspectral UAV images[J]. Remote Sensing, 2019, 11(13): 1554.
[8] HUANG Y N, QIAN Y R, WEI H Y, et al. A survey of deep learning-based object detection methods in crop counting[J]. Computers and Electronics in Agriculture, 2023, 215: 108425.
[9] MAO C Z, MENG W L, SHI C Y, et al. A crop disease image recognition algorithm based on feature extraction and image segmentation[J]. Traitement du Signal, 2020, 37(2): 341-346.
[10] LIU Z S, XIANG X Y, QIN J H, et al. Image recognition of citrus diseases based on deep learning[J]. Computers, Materials & Continua, 2020, 66(1): 457-466.
[11] APPE S N, ARULSELVI G, BALAJI G N. CAM-YOLO: tomato detection and classification based on improved YOLOv5 using combining attention mechanism[J]. PeerJ Computer Science, 2023, 9: e1463.
[12] 王会征, 孙良晨, 李新龙, 等. 基于改进YOLOv7-tiny的番茄叶片病虫害检测方法[J]. 农业工程学报, 2024, 40(10): 194-202.
WANG H Z, SUN L C, LI X L, et al. Detecting tomato leaf pests and diseases using improved YOLOv7-tiny[J]. Transactions of the Chinese Society of Agricultural Engineering, 2024, 40(10): 194-202.
[13] 李好, 邱卫根, 张立臣. 改进ShuffleNet V2的轻量级农作物病害识别方法[J]. 计算机工程与应用, 2022, 58(12): 260-268.
LI H, QIU W G, ZHANG L C. Improved ShuffleNet V2 for lightweight crop disease identification[J]. Computer Engineering and Applications, 2022, 58(12): 260-268.
[14] OLSEN A, KONOVALOV D A, PHILIPPA B, et al. DeepWeeds: a multiclass weed species image dataset for deep learning[J]. Scientific Reports, 2019, 9: 2058.
[15] 杨英茹, 吴华瑞, 张燕, 等. 基于复杂环境的番茄叶部图像病虫害识别[J]. 中国农机化学报, 2021, 42(9): 177-186.
YANG Y R, WU H R, ZHANG Y, et al. Tomato disease recognition using leaf image based on complex environment[J]. Journal of Chinese Agricultural Mechanization, 2021, 42(9): 177-186.
[16] BUTERA L, FERRANTE A, JERMINI M, et al. Precise agriculture: effective deep learning strategies to detect pest insects[J]. IEEE/CAA Journal of Automatica Sinica, 2022, 9(2): 246-258.
[17] 廖娟, 刘凯旋, 杨玉青, 等. 基于RDN-YOLO的自然环境下水稻病害识别模型研究[J]. 农业机械学报, 2024, 55(8): 233-242.
LIAO J, LIU K X, YANG Y Q, et al. Rice disease recognition in natural environment based on RDN-YOLO[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(8): 233-242.
[18] SHAO Y H, ZHANG D, CHU H Y, et al. A review of YOLO object detection based on deep learning[J]. Journal of Electronics & Information Technology, 2022, 44(10): 3697-3708.
[19] DING X H, ZHANG X Y, MA N N, et al. RepVGG: making VGG-style ConvNets great again[C]//Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2021: 13728-13737.
[20] CHEN Y F, ZHANG C Y, CHEN B, et al. Accurate leukocyte detection based on deformable-DETR and multi-level feature fusion for aiding diagnosis of blood diseases[J]. Computers in Biology and Medicine, 2024, 170: 107917.
[21] HUANG L, LI W S, TAN Y J, et al. YOLOCS: object detection based on dense channel compression for feature spatial solidification[J]. Knowledge-Based Systems, 2025, 310: 113024.
[22] ZHENG Z H, WANG P, REN D W, et al. Enhancing geometric factors in model learning and inference for object detection and instance segmentation[J]. IEEE Transactions on Cybernetics, 2022, 52(8): 8574-8586.
[23] 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. .
[24] 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.
[25] GEVORGYAN Z. SIoU loss: more powerful learning for bounding box regression[J]. arXiv:2205.12740, 2022.
[26] TONG Z J, CHEN Y H, XU Z W, et al. Wise-IoU: bounding box regression loss with dynamic focusing mechanism[J]. arXiv:2301.10051, 2023.
[27] CHEN P C, XU W C, ZHAN Y L, et al. Determining application volume of unmanned aerial spraying systems for cotton defoliation using remote sensing images[J]. Computers and Electronics in Agriculture, 2022, 196: 106912.