基于YOLOv8网络架构和无人机多光谱影像的柑橘黄化植株识别

doi:10.3969/j.issn.1004-1524.20240607

浙江农业学报 ›› 2026, Vol. 38 ›› Issue (8): 1674-1682.DOI: 10.3969/j.issn.1004-1524.20240607

基于YOLOv8网络架构和无人机多光谱影像的柑橘黄化植株识别

王宏乐¹(), 刘大存², 梁振伟²

¹ 深圳市五谷网络科技有限公司, 广东深圳 518000
² 深圳市丰农数智农业科技有限公司, 广东深圳 518063

收稿日期:2024-07-08 出版日期:2026-08-25 发布日期:2026-09-14
作者简介:王宏乐,研究方向为植物病理学和数字农业。E-mail: teajam@163.com
基金资助:
广东省现代农业产业园项目(GDSCYY2022-046);深圳市科技计划(CJGJZD20210408092401004);深圳市科技计划(KCXFZ20240903093800002)

Recognition of chlorotic citrus trees based on YOLOv8 network architecture with unmanned aerial vehicle multispectral images

WANG Hongle¹(), LIU Dacun², LIANG Zhenwei²

¹ Shenzhen Wego Network Co., Ltd., Shenzhen 518000, Guangdong, China
² Shenzhen SenseAgro Technology Co., Ltd., Shenzhen 518063, Guangdong, China

Received:2024-07-08 Published:2026-08-25 Online:2026-09-14

摘要/Abstract

摘要：

快速筛选黄化植株,有利于果园问题植株的精准定位与防控管理。基于无人机RGB图像的深度学习已被广泛应用于农业领域的快速检测,但基于YOLOv8网络架构的无人机多光谱图像识别报道较少。本研究建立了一种基于YOLOv8网络架构和无人机多光谱影像的柑橘植株识别分析方法,并通过特征波段筛选来进行方法优化和比较。结果表明:植株样本与背景的光谱特征有很大差异,而黄化植株与健康植株较难区分。基于无人机RGB影像建立的模型,对柑橘健康植株的识别效果较佳,但对柑橘黄化植株的识别效果较差。增加训练图片的光谱影像通道数量后, 模型对黄化植株的识别精度呈上升趋势。特征波段筛选对模型识别精度存在影响,当特征比重之和大于90%时,模型预测效果良好。本研究为进一步开发更稳定、高效,可大范围应用的柑橘黄化植株快速监测方法,和探索多光谱影像与深度学习结合的方法在农业领域的应用奠定了基础。

关键词: 柑橘, 黄化植株, 无人机, 多光谱, YOLOv8模型

Abstract:

Rapid screening of yellowed plants is beneficial for the precise positioning and control of unhealthy plants in orchards. Deep learning based on RGB images of unmanned aerial vehicle (UAV) has been widely applied for rapid detection in agriculture, but there are few reports about the deep learning based on multispectral images of UAV. This study proposed a detection and analytical framework for citrus trees based on YOLOv8 network architecturfe and UAV multispectral data, and further optimized and compared the model’s performance via characteristic band screening. The results showed that the features differed between citrus plants and background, but could not differ between healthy plants and chlorotic plants. Detection model of based on RGB images of UAV performed well on healthy citrus plants, but not well on chlorotic citrus plants. By increasing spectral images channels of training images, the recognition accuracy of chlorotic plants was improved. Bands selection had impact on the accuracy. When the cumulative proportion of feature was greater than 90%, the model performed well in prediction. This study provided a basis for developing a widely used, stable and efficient rapid monitoring method for citrus chlorotic plants, as well as advancing the integrated application of multispectral remote sensing and deep learning in precision agriculture.

Key words: citrus, chlorotic plant, unmanned aerial vehicle, multi-spectral, YOLOv8 model

中图分类号:

S24
TP391.41

王宏乐, 刘大存, 梁振伟. 基于YOLOv8网络架构和无人机多光谱影像的柑橘黄化植株识别[J]. 浙江农业学报, 2026, 38(8): 1674-1682.

WANG Hongle, LIU Dacun, LIANG Zhenwei. Recognition of chlorotic citrus trees based on YOLOv8 network architecture with unmanned aerial vehicle multispectral images[J]. Acta Agriculturae Zhejiangensis, 2026, 38(8): 1674-1682.

图/表 4

图1 训练集图片样例绿色框表示健康植株;红色框表示黄化植株。

Fig.1 Image samples of training set Green boxes show healthy plants, while red boxes show chlorotic plants.

图2 多光谱影像的特征提取(左)与主成分分析(右) G,健康植株;Y,黄化植株;B,背景。PC1和PC2分别为第1主成分和第2主成分。影像通道0~7分别表示R、G、B、B450(450 nm)、G560(560 nm)、R650(650 nm)、RE(730 nm)、NIR(840 nm)。

Fig.2 Feature extraction (left) and principal component analysis (right) of multi-spectral images G, Healthy plants; Y, Chlorotic plants; B, Background. PC1 and PC2 represent principal component 1 and 2, respectively. Channels of image 0-7 correspond to R, G, B, B450(450 nm), G560(560 nm), R650(650 nm), RE(730 nm) and NIR(840 nm) bands, respectively.

表1 多光谱影像在模型上的测试结果

Table 1 Test results of multispectral images based on the constructed model

处理 Treatment	验证集上的结果Results on validation set				测试集上的结果Results on test set
处理 Treatment	F1分数 F1 score	mAP@0.5_1	mAP@0.5_2	mAP@0.5_3	F1分数 F1 score	mAP@0.5_1	mAP@0.5_2	mAP@0.5_3
C3	0.77	0.841	0.969	0.714	0.74	0.783	0.866	0.699
C4	0.83	0.881	0.924	0.838	0.81	0.869	0.931	0.807
C5	0.81	0.845	0.905	0.785	0.80	0.860	0.898	0.822
C6	0.85	0.897	0.946	0.848	0.85	0.886	0.924	0.849
C7	0.80	0.883	0.899	0.866	0.83	0.901	0.923	0.879
C8	0.87	0.894	0.942	0.847	0.86	0.895	0.926	0.865
S3	0.68	0.689	0.926	0.452	0.72	0.758	0.920	0.596
S4	0.75	0.805	0.936	0.675	0.80	0.836	0.891	0.781
S5	0.85	0.897	0.922	0.872	0.84	0.888	0.905	0.870
S6	0.86	0.909	0.921	0.897	0.88	0.916	0.939	0.894
S7	0.84	0.900	0.971	0.829	0.90	0.932	0.943	0.920
S8	0.87	0.894	0.942	0.847	0.86	0.895	0.926	0.865

图3 YOLOv8模型对多通道影像的检测结果示例 a、b、c为不同波段随机组成的三通道图像检测结果示例。g,健康植株;y,黄化植株。

Fig.3 Examples of detection results of multi-channel images by YOLOv8 model a, b and c are examples of three random channels image detection results composed of different bands. g, Healthy plants; y, Chlorotic plants.

参考文献 37

[1]	林奕桐, 梁健, 刘书田, 等. 基于无人机可见光通道及支持向量机模型的柑橘黄龙病个体识别方法[J]. 西南农业学报, 2022, 35(11): 2554-2563.
	Lin Y T, Liang J, Liu S T, et al. Individual identification method of citrus Huanglongbing based on UAV visible light channel and support vector machine model[J]. Southwest China Journal of Agricultural Sciences, 2022, 35(11): 2554-2563.
[2]	束美艳, 李世林, 魏家玺, 等. 基于无人机平台的柑橘树冠信息提取[J]. 农业工程学报, 2021, 37(1): 68-76.
	Shu M Y, Li S L, Wei J X, et al. Extraction of citrus crown parameters using UAV platform[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(1): 68-76.
[3]	刘建刚, 赵春江, 杨贵军, 等. 无人机遥感解析田间作物表型信息研究进展[J]. 农业工程学报, 2016, 32(24): 98-106.
	Liu J G, Zhao C J, Yang G J, et al. Review of field-based phenotyping by unmanned aerial vehicle remote sensing platform[J]. Transactions of the Chinese Society of Agricultural Engineering, 2016, 32(24): 98-106.
[4]	宋昊昕, 尤号田, 刘遥, 等. 基于深度学习的无人机多光谱图像柑橘树冠分割[J]. 森林工程, 2023, 39(3): 140-149.
	Song H X, You H T, Liu Y, et al. Deep learning-based segmentation of citrus tree canopy from UAV multispectral images[J]. Forest Engineering, 2023, 39(3): 140-149.
[5]	陈周娟, 程光, 卜元坤, 等. 基于无人机倾斜影像的阔叶林单木参数提取[J]. 林业资源管理, 2022(1): 132-141.
	Chen Z J, Cheng G, Bu Y K, et al. Single tree parameters extraction of broad-leaved forest based on UAV tilting photography[J]. Forest Resources Management, 2022(1): 132-141.
[6]	邓家锐, 敖义俊, 余小丽, 等. 城固柑橘黄化现象成因调查分析与对策建议[J]. 陕西农业科学, 2023, 69(2): 105-108.
	Deng J R, Ao Y J, Yu X L, et al. Investigation and analysis on cause of citrus yellowing in Chenggu and its countermeasures[J]. Shaanxi Journal of Agricultural Sciences, 2023, 69(2): 105-108.
[7]	宋勇, 陈兵, 王琼, 等. 基于无人机多光谱影像的棉花黄萎病监测[J]. 棉花学报, 2023, 35(2): 87-100.
	Song Y, Chen B, Wang Q, et al. Monitoring of cotton Verticillium wilt based on unmanned aerial vehicle multispectral images[J]. Cotton Science, 2023, 35(2): 87-100.
[8]	刘浪, 许金钗, 翁海勇, 等. 基于多光谱成像技术的水稻叶部病害检测装置设计与试验[J]. 福建农林大学学报(自然科学版), 2023, 52(2): 280-284.
	Liu L, Xu J C, Weng H Y, et al. Development of a portable detection device for rice leaf diseases using multispectral imaging[J]. Journal of Fujian Agriculture and Forestry University(Natural Science Edition), 2023, 52(2): 280-284.
[9]	兰玉彬, 朱梓豪, 邓小玲, 等. 基于无人机高光谱遥感的柑橘黄龙病植株的监测与分类[J]. 农业工程学报, 2019, 35(3): 92-100.
	Lan Y B, Zhu Z H, Deng X L, et al. Monitoring and classification of citrus Huanglongbing based on UAV hyperspectral remote sensing[J]. Transactions of the Chinese Society of Agricultural Engineering, 2019, 35(3): 92-100.
[10]	兰玉彬, 王天伟, 郭雅琦, 等. 柑橘黄龙病光谱特征波段选择及光谱检测仪研制[J]. 农业工程学报, 2022, 38(20): 119-128.
	Lan Y B, Wang T W, Guo Y Q, et al. Selection of spectral characteristic bands of HLB disease of citrus and spectrum detector development[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(20): 119-128.
[11]	李敏, 覃泽林, 兰宗宝, 等. 基于无人机低空高光谱遥感影像的柑橘黄龙病植株监测模型[J]. 湖北农业科学, 2023, 62(8): 207-212.
	Li M, Qin Z L, Lan Z B, et al. Citrus huanglongbing plant monitoring model based on UAV low altitude hyperspectral remote sensing imaging[J]. Hubei Agricultural Sciences, 2023, 62(8): 207-212.
[12]	吴利刚, 陈乐, 刘泽鹏, 等. 基于YOLOv8-ABW的黄花成熟度检测方法[J]. 农业工程学报, 2024, 40(13): 262-272.
	Wu L G, Chen L, Liu Z P, et al. YOLOv8-ABW based method for detecting Hemerocallis citrina Baroni maturity[J]. Transactions of the Chinese Society of Agricultural Engineering, 2024, 40(13): 262-272.
[13]	宁孝梅, 马慧敏, 王小申, 等. 基于深度学习的种子质量检测综述[J]. 河南工业大学学报(自然科学版), 2024, 45(2): 140-149.
	Ning X M, Ma H M, Wang X S, et al. A review of seed quality detection based on deep learning[J]. Journal of Henan University of Technology(Natural Science Edition), 2024, 45(2): 140-149.
[14]	黄硕, 周亚男, 王起帆, 等. 改进YOLOv5测量田间小麦单位面积穗数[J]. 农业工程学报, 2022, 38(16): 235-242.
	Huang S, Zhou Y N, Wang Q F, et al. Measuring the number of wheat spikes per unit area in fields using an improved YOLOv5[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(16): 235-242.
[15]	鲍文霞, 谢文杰, 胡根生, 等. 基于TPH-YOLO的无人机图像麦穗计数方法[J]. 农业工程学报, 2023, 39(1): 155-161.
	Bao W X, Xie W J, Hu G S, et al. Wheat ear counting method in UAV images based on TPH-YOLO[J]. Transactions of the Chinese Society of Agricultural Engineering, 2023, 39(1): 155-161.
[16]	臧贺藏, 赵晴, 周萌, 等. 基于YOLOv5s模型的小麦品种(系)穗数检测[J]. 山东农业科学, 2022, 54(11): 150-157.
	Zang H C, Zhao Q, Zhou M, et al. Spike number detection of wheat varieties (lines) based on YOLOv5s model[J]. Shandong Agricultural Sciences, 2022, 54(11): 150-157.
[17]	郭瑞, 于翀宇, 贺红, 等. 采用改进YOLOv4算法的大豆单株豆荚数检测方法[J]. 农业工程学报, 2021, 37(18): 179-187.
	Guo R, Yu C Y, He H, et al. Detection method of soybean pod number per plant using improved YOLOv4 algorithm[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(18): 179-187.
[18]	陈优良, 张向君, 陈勋俊. 深度学习算法YOLOv4支持下的脐橙树株数识别[J]. 测绘科学, 2022, 47(2): 135-144.
	Chen Y L, Zhang X J, Chen X J. Identification of navel orange trees based on deep learning algorithm YOLOv4[J]. Science of Surveying and Mapping, 2022, 47(2): 135-144.
[19]	Gani M O, Kuiry S, Das A, et al. Multispectral object detection with deep learning[C]//Computational intelligence in communications and business analytics. Cham: Springer, 2021: 105-117.
[20]	Wang S, Zeng D W, Xu Y X, et al. Towards complex scenes: a deep learning-based camouflaged people detection method for snapshot multispectral images[J]. Defence Technology, 2024, 34: 269-281.
[21]	Varnima E K, Ramachandran C. Real-time gender identification from face images using You Only Look Once (YOLO)[C]// 2020 4th International Conference on Trends in Electronics and Informatics (ICOEI), 2020: 1074-1077.
[22]	杨禄, 蒋晓华, 常文涛, 等. 基于“珠海一号”高光谱遥感影像的薇甘菊入侵监测[J]. 亚热带植物科学, 2022, 51(4): 272-281.
	Yang L, Jiang X H, Chang W T, et al. Monitoring of Mikania micrantha invasion based on ‘Zhuhai-1’ hyperspectral remote sensing images[J]. Subtropical Plant Science, 2022, 51(4): 272-281.
[23]	王勇, 曾体伟, 徐秋, 等. 基于无人机多光谱图像和集成学习的橡胶树白粉病检测[J]. 石河子大学学报(自然科学版), 2024, 42(3): 265-274.
	Wang Y, Zeng T W, Xu Q, et al. Detection of rubber tree powdery mildew based on UAV multispectral image and stacked machine learning model[J]. Journal of Shihezi University(Natural Science), 2024, 42(3): 265-274.
[24]	李安琪, 杨琳, 蔡言颜, 等. 基于递归特征消除-随机森林模型的江浙沪农田土壤肥力属性制图[J]. 地理科学, 2024, 44(1): 168-178.
	Li A Q, Yang L, Cai Y Y, et al. Digital mapping of soil fertility attributes in croplands in Jiangsu, Zhejiang and Shanghai based on recursive feature elimination-random forest model[J]. Scientia Geographica Sinica, 2024, 44(1): 168-178.
[25]	王宏乐, 王兴林, 李文波, 等. 一种基于解耦旋转锚框匹配策略的谷粒检测方法[J]. 广东农业科学, 2022, 49(12): 143-150.
	WANG H L, WANG X L, LI W B, et al. A rice grain detection method based on rotation-decoupled detector for oriented object[J]. Guangdong Agricultural Sciences, 2022, 49(12): 143-150.
[26]	周思捷, 刘天奇, 陈天华. 基于改进YOLOv5算法的水稻病害识别研究[J]. 中国农机化学报, 2024, 45(8): 246-253.
	Zhou S J, Liu T Q, Chen T H. Research on rice disease recognition based on improved YOLOv5 algorithm[J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 246-253.
[27]	Loshchilov I, Hutter F. SGDR: stochastic gradient descent with warm restarts[PP/OL]. arXiv (2016-08-13) [2024-07-08]. https://arxiv.org/abs/1608.03983.
[28]	Gotmare A, Keskar N S, Xiong C M, et al. A closer look at deep learning heuristics: learning rate restarts, warmup and distillation[PP/OL]. arXiv (2018-10-29) [2024-07-08]. https://arxiv.org/abs/1810.13243.
[29]	魏靖, 王玉亭, 袁会珠, 等. 基于深度学习与特征可视化方法的草地贪夜蛾及其近缘种成虫识别[J]. 智慧农业(中英文), 2020, 2(3): 75-85.
	Wei J, Wang Y T, Yuan H Z, et al. Identification and morphological analysis of adult Spodoptera frugiperda and its close related species using deep learning[J]. Smart Agriculture, 2020, 2(3): 75-85.
[30]	Sapkota R, Ahmed D, Karkee M. Comparing YOLOv8 and Mask R-CNN for instance segmentation in complex orchard environments[J]. Artificial Intelligence in Agriculture, 2024, 13: 84-99.
[31]	Sharma A, Kumar V, Longchamps L. Comparative performance of YOLOv8, YOLOv9, YOLOv10, YOLOv11 and Faster R-CNN models for detection of multiple weed species[J]. Smart Agricultural Technology, 2024, 9: 100648.
[32]	窦世卿, 张晨红, 靖娟利, 等. 融合高斯白噪声的PROSPECT模型的柑橘叶片叶绿素反演[J]. 无线电工程, 2024, 54(7): 1702-1711.
	Dou S Q, Zhang C H, Jing J L, et al. Chlorophyll inversion of citrus leaves in PROSPECT model with Gaussian white noise[J]. Radio Engineering, 2024, 54(7): 1702-1711.
[33]	Broge N H, Mortensen J V. Deriving green crop area index and canopy chlorophyll density of winter wheat from spectral reflectance data[J]. Remote Sensing of Environment, 2002, 81(1): 45-57.
[34]	张春琪, 刘梦姝, 晁金阳, 等. 基于特征波长的接触式作物叶绿素检测系统[J]. 农业机械学报, 2024, 55(S2): 255-262.
	Zhang C Q, Liu M S, Chao J Y, et al. Contact-based crop chlorophyll detection system based on feature wavelengths[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(S2): 255-262.
[35]	Zheng Z J, Yuan J H, Yao W, et al. Fusion of UAV-acquired visible images and multispectral data by applying machine-learning methods in crop classification[J]. Agronomy, 2024, 14(11): 2670.
[36]	Galymzhankyzy Z, Martinson E. Lightweight multispectral crop-weed segmentation for precision agriculture[PP/OL]. arXiv (2025-05-12) [2026-04-03]. https://arxiv.org/abs/2505.07444.
[37]	Zhang L L, Zhang B, Zhang H H, et al. Multi-source feature fusion network for LAI estimation from UAV multispectral imagery[J]. Agronomy, 2025, 15(4): 988.

基于YOLOv8网络架构和无人机多光谱影像的柑橘黄化植株识别

Recognition of chlorotic citrus trees based on YOLOv8 network architecture with unmanned aerial vehicle multispectral images

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 4

参考文献 37

相关文章 15

编辑推荐

Metrics

本文评价

[1]	温明霞, 李永杰, 黄贝. 不同绿肥管理模式对柑橘园土壤理化特性和细菌群落的调控效应[J]. 浙江农业学报, 2026, 38(7): 1432-1442.
[2]	曹梦娇, 白石, 唐攀攀, 王晔青, 徐红星, 周国鑫. 基于无人机多光谱遥感的早稻二化螟为害程度评价[J]. 浙江农业学报, 2025, 37(3): 603-611.
[3]	沈佳瑜, 邓燕, 张姬雯, 彭国方, 吴群, 朱齐超, 张卫峰, 李贝. 衢州市柑橘园碳排放与碳储量现状评价[J]. 浙江农业学报, 2025, 37(11): 2340-2353.
[4]	李萌民, 刘朔, 欧阳宇, 张鹏. 基于YOLOv8n的高效轻量化柑橘叶片病害检测模型[J]. 浙江农业学报, 2025, 37(10): 2198-2208.
[5]	程嘉瑜, 陈妙金, 李彤, 孙奇男, 张小斌, 赵懿滢, 朱怡航, 顾清. 基于改进Faster-RCNN网络的无人机遥感影像桃树检测[J]. 浙江农业学报, 2024, 36(8): 1909-1919.
[6]	李大华, 孔舒, 李栋, 于晓. 基于改进SSD模型的柑橘叶片病害轻量化检测模型[J]. 浙江农业学报, 2024, 36(3): 662-670.
[7]	郭发旭, 冯全, 杨森, 杨婉霞. 基于无人机高光谱的马铃薯冠层叶片全氮含量反演[J]. 浙江农业学报, 2023, 35(8): 1904-1914.
[8]	张梦, 佘宝, 杨玉莹, 黄林生, 朱梦琦. 基于无人机RGB影像的大豆种植区提取方法研究[J]. 浙江农业学报, 2023, 35(4): 952-961.
[9]	朱永基, 陶新宇, 陈小芳, 苏祥祥, 刘吉凯, 李新伟. 基于无人机多光谱影像植被指数与纹理特征的冬小麦地上部生物量估算[J]. 浙江农业学报, 2023, 35(12): 2966-2976.
[10]	姜友谊, 张成健, 韩少宇, 杨小冬, 杨贵军, 杨浩. 基于无人机三维点云的玉米植株自动计数研究[J]. 浙江农业学报, 2022, 34(9): 2032-2042.
[11]	曾雅婷, 熊桃, 李红叶. 柑橘黑点病菌(Diaporthe citri)快速分子检测技术[J]. 浙江农业学报, 2022, 34(7): 1457-1465.
[12]	王钧, 陆洲, 罗明, 徐飞飞, 张序. 基于机载多光谱的冬小麦返青期土壤墒情反演[J]. 浙江农业学报, 2022, 34(6): 1297-1305.
[13]	习文勇, 傅佩红. 基于模糊数学的柑橘种植土地适宜性评价[J]. 浙江农业学报, 2022, 34(1): 141-152.
[14]	邬宁珊, 王佳希, 张岩, 元慕田, 张琪, 高驰宇. 基于无人机可见光影像的树种和树冠信息提取——以晋西黄土区蔡家川流域为例[J]. 浙江农业学报, 2021, 33(8): 1505-1518.
[15]	孟幼青, 汪恩国, 陈吴健, 李艳敏, 程帆, 孟敏霞. 浙江省亚洲柑橘木虱黄龙病病原分布与消长规律[J]. 浙江农业学报, 2021, 33(3): 464-469.