基于剪枝和知识蒸馏的YOLOv8轻量化苹果叶片病害检测方法

张帅平; 时雷; 郑光; 王慧新; 尹飞

doi:10.13300/j.cnki.hnlkxb.2026.03.009

您当前的位置：

首页 >

文章列表页 >

基于剪枝和知识蒸馏的YOLOv8轻量化苹果叶片病害检测方法

农业人工智能专题 | 更新时间：2026-11-05

- 基于剪枝和知识蒸馏的YOLOv8轻量化苹果叶片病害检测方法
- A pruning and knowledge distillation-based YOLOv8 lightweight method for detecting leaf diseases in apple
- 华中农业大学学报 2026年45卷第3期页码：98-114
- 作者机构：
  
  河南农业大学信息与管理科学学院，郑州 450046
- 作者简介：
  
  张帅平， E-mail： zsp2292@stu.henau.edu.cn
  尹飞， E-mail： yin.fei@henau.edu.cn
- 基金信息：
  
  河南省科技攻关项目(242102521027);河南省科技研发计划联合基金项目(222301420113)
- DOI：10.13300/j.cnki.hnlkxb.2026.03.009
  中图分类号： TP391.4;S661.1
- 收稿：2025-09-25，
  
  纸质出版：2026-05-25
- 稿件说明：
移动端阅览
张帅平，时雷，郑光，等.基于剪枝和知识蒸馏的YOLOv8轻量化苹果叶片病害检测方法［J］.华中农业大学学报，2026，45（3）：98‐114.

ZHANG Shuaiping,SHI Lei,ZHENG Guang,et al.A pruning and knowledge distillation-based YOLOv8 lightweight method for detecting leaf diseases in apple[J].Journal of Huazhong Agricultural University,2026,45(03):98-114.
张帅平，时雷，郑光，等.基于剪枝和知识蒸馏的YOLOv8轻量化苹果叶片病害检测方法［J］.华中农业大学学报，2026，45（3）：98‐114. DOI： 10.13300/j.cnki.hnlkxb.2026.03.009.

ZHANG Shuaiping,SHI Lei,ZHENG Guang,et al.A pruning and knowledge distillation-based YOLOv8 lightweight method for detecting leaf diseases in apple[J].Journal of Huazhong Agricultural University,2026,45(03):98-114. DOI： 10.13300/j.cnki.hnlkxb.2026.03.009.

摘要

为解决现有苹果叶片检测方法在准确率、实时性及复杂噪声环境下鲁棒性等方面的问题，基于YOLOv8n提出一种轻量化、实时苹果叶片病害检测模型SMPD-YOLO。首先，将SPPFCSPC（spatial pyramid pooling fast cross stage partial CSP）特征金字塔模块嵌入到主干网络，增强特征融合能力。其次，引入MPD-IoU（minimum point distance-IoU）作为边界框回归损失函数，提高模型的精度和收敛速度。再次，通过基于层自适应幅度剪枝（layer adaptive magnitude-based pruning，LAMP）进一步压缩模型体积、降低浮点运算速度。最后，采用通道级知识蒸馏（channel-wise knowledge distillation，CWD）策略，提升检测性能。实验结果显示，改进后SMPD-YOLO模型在IoU（intersection over union）阈值大于0.5时的平均精度均值和帧率分别达到90.20%和133.3帧/s，模型权重、浮点运算速度分别为5.0 MB、7.3×10

-1

。此外，改进模型在强光、弱光及图像模糊复杂噪声环境下仍能展现出优异的鲁棒性。结果表明，SMPD-YOLO模型兼具高准确性、轻量化和实时性，可在资源受限设备上实现高效叶片病害检测。

Abstract

A lightweight， real-time model of detecting leaf diseases in apple， named as SMPD-YOLO， was proposed based on YOLOv8n to solve the problems in existing methods of detecting leaf diseases in apple in terms of accuracy， real-time performance， and robustness under complex and noisy environments. The spatial pyramid pooling fast cross stage partial CSP （SPPFCSPC） module was embedded into the backbone network t

o enhance the capability of feature fusion. The minimum point distance-IoU （MPD-IoU） was introduced as the bounding box regression loss function to improve the accuracy and convergence speed of the model. The model volume and floating point operations were further reduced via layer adaptive magnitude‑based pruning （LAMP）. A channel-wise knowledge distillation （CWD） strategy was used to boost detection performance. The results showed that the improved SMPD-YOLO model had a mean average precision（mAP@0.5） of 90.20% and a frame rate of 133.3 frames per second （FPS），while the weight and FLOPs of model was 5.0 MB and 7.3×10

-1

， respectively. The improved model maintained excellent robustness under complex and noisy environments including strong illumination， weak light， and blurred images. It is indicated that the SMPD-YOLO model combines high accuracy， lightweight design， and real-time performance， enabling the efficient detection of leaf diseases on resource-constrained equipment.

关键词

Keywords

references

SANGEETHA K，VISHNU RAJA P，RIMA P，et al.Apple leaf diseas e detection using deep learning［C］//2022 6th International Conference on Computing Methodologies and Communication （ICCMC）. March 29-31，2022 ，Erode，India.Erode：IEEE，2022：1063-1067.

KHAN A I，QUADRI S M K，BANDAY S，et al.Deep diagnosis：a real-time apple leaf disease detection system based on deep learning［J/OL］.Computers and electronics in agriculture，2022，198：107093［2025-09-25］.https：//doi.org/10.1016/j.compag.2022.107093.

LI J W，QIAO Y L，LIU S，et al.An improved YOLOv5-based vegetable disease detection method［J/OL］.Computers and electronics in agriculture，2022，202：107345［2025-09-25］.https：//doi.org/10.1016/j.compag.2022.107345.

ZHANG S W，WANG D W，YU C Q.Apple leaf disease recognition method based on Siamese dilated Inception network with less training samples［J/OL］.Computers and electronics in agriculture，2023，213：108188［2025-09-25］.https：//doi.org/10.1016/j.compag.2023.108188.

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.

姜晟，曹亚芃，刘梓伊，等.基于改进Faster RCNN的茶叶叶部病害识别［J］.华中农业大学学报，2024，43（5）：41-50. JIANG S，CAO Y P，LIU Z Y，et al.Recognition of tea leaf disease based on improved Faster RCNN［J］.Journal of Huazhong Agricultural University，2024，43（5）：41-50 （in Chinese with English abstract） .

LI Z S，SHEN Y，TANG J S，et al.IMLL-DETR：an intelligent model for detecting multi-scale litchi leaf diseases and pests in complex agricultural environments［J/OL］.Expert systems with applications，2025，273：126816［2025-09-25］. https：//doi.org/10.1016/j.eswa.2025.126816.

YAO J B，ZHU Z H，YUAN M Q，et al.The detection of maize leaf disease based on an improved real-time detection transformer model［J/OL］.Symmetry，2025，17（6）：808［2025-09-25］.https：//doi.org/10.3390/sym17060808.

赵嘉威，田光兆，邱畅，等.基于改进YOLOv4算法的苹果叶片病害检测方法［J］.江苏农业科学，2023，51（9）：193-199.ZHAO J W，TIAN G Z，QIU C，et al.Apple leaf disease detection method based on improved YOLOv4 algorithm ［J］.Jiangsu agricultural sciences，2023，51（9）：193-199 （in Chinese with English abstract） .

李昱达，吴正平，孙水发，等.改进YOLOv5算法的多类苹果叶片病害检测［J］.中国农机化学报，2024，45（12）：230-237. LI Y D，WU Z P，SUN S F，et al.Multi-species apple leaf disease detection with improved YOLOv5 algorithm［J］.Journal of Chinese agricultural mechanization，2024，45（12）：230-237 （in Chinese with English abstract） .

袁杰，谢霖伟，郭旭，等.基于改进YOLOv7的苹果叶片病害检测方法［J］.农业机械学报，2024，55（11）：68-74. YUAN J，XIE L W，GUO X，et al.Apple leaf disease detection method based on improved YOLOv7［J］.Transactions of the CSAM，2024，55（11）：68-74 （in Chinese with English abstract） .

张惠莉，代晨龙，任景龙，等.基于YOLOv8-GSGF模型的葡萄病害识别方法研究［J］.农业机械学报，2024，55（11）：75-83. ZHANG H L，DAI C L，REN J L，et al.Grape disease identification method based on YOLOv8-GSGF model［J］.Transactions of the CSAM，2024，55（11）：75-83 （in Chinese with English abstract） .

曾晏林，贺壹婷，蔺瑶，等. 基于BCE-YOLOv5的苹果叶部病害检测方法［J］. 江苏农业科学， 2023， 51 （15）： 155-163.ZENG Y L， HE Y T， LIN Y， et al. Apple leaf disease detection method based on BCE-YOLOv5 ［J］. Jiangsu agricultural sciences， 2023， 51（15）： 155-163 （in Chinese with English abstract） .

黄政，张涛，孔万仔，等.基于YOLOv8n的梨树叶片病害检测模型［J］.湖南农业大学学报（自然科学版），2025，51（2）：113-121. HUANG Z，ZHANG T，KONG W Z，et al.Detection model of pear leaf disease based on YOLOv8n［J］.Journal of Hunan Agricultural University （natural sciences edition），2025，51（2）：113-121 （in Chinese with English abstract） .

XUE Z Y，XU R J，BAI D，et al.YOLO-tea：a tea disease detection model improved by YOLOv5［J/OL］.Forests，2023，14（2）：415［2025-09-25］.https：//doi.org/10.3390/f14020415.

THAPA R，ZHANG K，SNAVELY N，et al.The plant pathology challenge 2020 data set to classify foliar disease of apples［J/OL］.Applications in plant sciences，2020，8（9）：e11390［2025-09-25］.https：//doi.org/10.1002/aps3.11390.

THAPA R， ZHANG K， SNAVELY N， et al. Plant pathology 2021-FGVC8［EB/OL］. ［2025-09-25］. https：//kaggle.com/competitions/plant-pathology-2021-fgvc8.

SINGH D，JAIN N，JAIN P，et al.PlantDoc：a dataset for visual plant disease detection［C］//Proceedings of the 7th ACM IKDD CoDS and 25th COMAD. January 5-7，2020 ，Hyderabad India.New York：ACM，2020：249-253.

HUSSAIN M.YOLO-v1 to YOLO-v8，the rise of YOLO and its complementary nature toward digital manufacturing and industrial defect detection［J/OL］.Machines，2023，11（7）：677［2025-09-25］.https：//doi.org/10.3390/machines11070677.

LIU L Z，LI P R，WANG D H，et al.A wind turbine damage detection algorithm designed based on YOLOv8［J/OL］.Applied soft computing，2024，154：111364［2025-09-25］. https：//doi.org/10.1016/j.asoc.2024.111364.

SANDHYA，KASHYAP A.A novel method for real-time object-based copy-move tampering localization in videos using fine-tuned YOLO V8［J/OL］.Forensic science international：digital investigation，2024，48：301663［2025-09-25］.https：//doi.org/10.1016/j.fsidi.2023.301663.

LIN T Y， DOLLÁR P， GIRSHICK R， et al. Feature pyramid networks for object detection［C］// 2017 IEEE Conference on Computer Vision and Pattern Recognition （CVPR）. July 21-26，2017 ，Honolulu， HI， USA. Honolulu：IEEE，2017： 2117-2125.

LIU S，QI L，QIN H F，et al.Path aggregation network for instance segmentation［PP/OL］.arXiv，2018：1803.01534［2025-09-25］.https：//doi.org/10.48550/arXiv.1803.01534.

WANG X Q，GAO H B，JIA Z M，et al.BL-YOLOv8：an improved road defect detection model based on YOLOv8［J/OL］.Sensors，2023，23（20）：8361［2025-09-25］.https：//doi.org/10.3390/s23208361.

WU T Y，DONG Y K.YOLO-SE：improved YOLOv8 for remote sensing object detection and recognition［J/OL］.Applied sciences，2023，13（24）：12977［2025-09-25］.https：//doi.org/10.3390/app132412977.

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.

TERVEN J，CÓRDOVA-ESPARZA D M，ROMERO-GONZÁLEZ J A.A comprehensive review of YOLO architectures in computer vision：from YOLOv1 to YOLOv8 and YOLO-NAS［J］.Machine learning and knowledge extraction，2023，5（4）：1680-1716.

CHEN L C，PAPANDREOU G，KOKKINOS I，et al.DeepLab：semantic image segmentation with deep convolutional nets，atrous convolution，and fully connected CRFs［J］.IEEE transactions on pattern analysis and machine intelligence，2018，40（4）：834-848.

WANG C Y，BOCHKOVSKIY A，LIAO H M.YOLOv7：trainable bag-of-freebies sets new state-of-the-art for real-time object detectors［C］//2023 IEEE/CVF Conferenc e on Computer Vision and Pattern Recognition （CVPR）. June 17-24，2023 ，Vancouver，BC，Canada.Vancouver：IEEE，2023：7464-7475.

YAN J H，ZHOU Z，ZHOU D J，et al.Underwater object detection algorithm based on attention mechanism and cross-stage partial fast spatial pyramidal pooling［J/OL］.Frontiers in marine science，2022，9：1056300［2025-09-25］.https：//doi.org/10.3389/fmars.2022.1056300.

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.

MA S L，XU Y.MPDIoU：a loss for efficient and accurate bounding box regression［PP/OL］.arXiv，2023：2307.07662［2025-09-25］.https：//arxiv.org/abs/2307.07662.

GALE T，ELSEN E，HOOKER S.The state of sparsity in deep neural networks［PP/OL］.arXiv， 2019 ：1902.09574［2025-09-25］.https：//arxiv.org/abs/1902.09574.

GALE T，ZAHARIA M，YOUNG C，et al.Sparse GPU kernels for deep learning［C］//SC20：International Conference for High Performance Computing，Networking，Storage and Analysis. November 9-19，2020 ，Atlanta，GA，USA.Atlanta：IEEE，2020：1-14.

LEE J，PARK S，MO S，et al.Layer-adaptive sparsity for the magnitude-bas ed pruning［PP/OL］.arXiv， 2020 ：2010.07611［2025-09-25］.https：//arxiv.org/abs/2010.07611.

ROMERO A，BALLAS N，KAHOU S E，et al.FitNets：hints for thin deep nets［PP/OL］.arXiv，2014：1412.6550［2025-09-25］.https：//arxiv.org/abs/1412.6550.

SHU C Y，LIU Y F，GAO J F，et al.Channel-wise knowledge distillation for dense prediction［C］//2021 IEEE/CVF International Conference on Computer Vision （ICCV）. October 10-17，2021 ，Montreal，QC，Canada.Montreal：IEEE，2021：5291-5300.

TONG Z J，CHEN Y H，XU Z W，et al.Wise-IoU：bounding box regression loss with dynamic focusing mechanism［PP/OL］.arXiv，2023：2301.10051［2025-09-25］.https：//arxiv.org/abs/2301.10051.

ZHANG H，ZHANG S J.Focaler-IoU：more focused intersection over union loss［PP/OL］.arXiv，2024：2401.10525［2025-09-25］.https：//arxiv.org/abs/2401.10525.

YUE K，DENG J，ZHOU F. Matching guided distillation［PP/OL］.arXiv， 2020 ：2008.09958［2025-09-25］. https：//doi.org/10.48550/arXiv.2008.09958.

WEI Y，PAN X，QIN H， et al. Quantization mimic：towards very tiny CNN for object detection［C］//European Conference on Computer Vision （ECCV）2018. September 8-14，2018 ，Munich， Germany. Cham：Springer，2018：267-283.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于YOLOv8n改进的玉米幼苗杂草识别模型

基于EDW-YOLOv8的棉花叶片病害检测

EMD-YOLO：基于YOLOv8n改进的轻量化作物叶部病害检测算法

畜禽养殖废弃物资源化利用技术及环境评估研究进展

大豆灰斑病病原变异、抗性挖掘与智能防控