地黄叶片虫害孔洞图像数据集（RPHD）的构建与基准评估

doi:10.19788/j.issn.2096-6369.100072

摘要/Abstract

摘要：

地黄是我国重要的道地中药材，规模化种植的地黄常受多种害虫侵袭，导致叶片孔洞频现，严重影响地黄的产量与品质。然而，当前基于深度学习的虫害孔洞检测方法面临田间环境复杂、目标尺度小、形态不规则等挑战。本研究构建了面向田间复杂环境的地黄叶片虫害孔洞图像数据集（Rehmannia Pest-induced Hole Dataset, RPHD）。该数据集包含初始采集版本（298张图像，5 059个标注）与精校增强版本（291张图像，2 678个高质量标注）。通过系统化的图像预处理流程，包括基于内容感知的裁剪与分辨率统一（1024×1024）、自适应双边滤波去噪与光照均衡化，并结合精细化人工标注与多轮交叉校验质量控制机制，显著提升了数据集的目标判别性。实验采用YOLOv10n、YOLOv11n和YOLO12n三种主流目标检测模型对数据集进行基准评估，对比两个版本在不同模型上的性能表现。结果表明，在三个模型上，使用精校增强数据集训练均取得显著性能提升：YOLOv10n的mAP@0.5从40.3%提升至87.2%，YOLOv11n从48.9%提升至92.1%，YOLO12n从51.2%提升至93.0%。该数据集为地黄小目标病虫害检测算法研究提供了标准化、高质量的基准数据支持，可供其他农作物相关研究借鉴，有助于推动农业视觉检测技术的实用化与智能化发展。

数据摘要：

项目	内容
数据集名称	地黄叶片虫害孔洞图像数据集（RPHD）
所属学科	计算机科学；农业科学
研究主题	虫害孔洞检测，图像数据集，目标检测，地黄
数据时间范围	2024年6月—2024年9月
时间分辨率	不适用
数据地理空间覆盖	河南省温县核心种植区（东经112.95°－113.15°，北纬34.88°－35.00°）
空间分辨率	不适用
数据类型与技术格式	原始田间图像（JPEG格式）；预处理后图像（JPEG格式，1024×1024像素）；目标检测标注文件（YOLO格式.txt）
数据库（集）组成	原始版本（298张图像，5 059个标注）；优化版本（291张图像，2 678个标注）
数据量	约848 MB
主要数据指标	虫害孔洞边界框坐标（x_center, y_center, width, height），类别ID=0
数据可用性	DOI:10.57760/sciencedb.29865; https://www.scidb.cn/detail?dataSetId=8954e40ec0eb4f3fb7dad889e982547f CSTR: 31253.11.sciencedb.29865; https://www.scidb.cn/s/RF7J3q

关键词: 地黄, 虫害孔洞检测, 图像数据集, 数据标注, YOLO, 农业人工智能

Abstract:

Rehmannia glutinosa is an important traditional Chinese medicinal herb. Its large-scale cultivation is often affected by various pests, leading to frequent leaf pest-induced holes that severely impact yield and quality. However, current deep learning-based Pest-induced Hole detection methods face challenges such as complex field environments, small target scales, and irregular morphological characteristics. To address these issues, this paper presents the first Rehmannia Leaf Pest-induced Hole Dataset (RPHD) specifically designed for complex field environments. The dataset comprises an initial version (298 images with 5,059 annotations) and a refined version (291 images with 2,678 high-quality annotations). Through a systematic image preprocessing pipeline, including content-aware cropping and resolution normalization (1024×1024), adaptive bilateral filtering for noise reduction, and illumination equalization, combined with meticulous manual annotation and multiple rounds of cross-validation quality control mechanisms, the annotation consistency and target discriminability of the dataset were significantly improved. Three mainstream object detection models—YOLOv10n, YOLOv11n, and YOLOv12n—were employed to evaluate the dataset, comparing the performance of the two versions across different models. Experimental results show that training on the refined dataset yields substantial performance improvements across all models:YOLOv10n: mAP@0.5 increased from 40.3% to 87.2%;YOLOv11n: mAP@0.5 increased from 48.9% to 92.1%;YOLO12n: mAP@0.5 increased from 51.2% to 93.0%. This dataset provides standardized, high-quality benchmark data for research on small-target pest and disease detection algorithms for Rehmannia and other crops, contributing to the practical and intelligent development of agricultural vision detection technologies.

Data summary:

Items	Description
Dataset name	Rehmannia Leaf Pest-induced Hole Image Dataset (RPHD)
Specific subject area	Computer science; Agricultural science
Research topic	Pest-induced hole detection, Image dataset, Object detection, Rehmannia glutinosa
Time range	June 2024 - September 2024
Temporal resolution	Not applicable
Geographical scope	Core planting area of Wen County, Henan Province, China (112.95°E-113.15°E, 34.88°N-35.00°N)
Spatial resolution	Not applicable
Data types and technical formats	Raw field images (JPEG); Preprocessed images (JPEG, 1024×1024 pixels); Object detection annotation files (YOLO format.txt)
Dataset structure	Raw version: 298 images, 5,059 annotations; Enhanced version: 291 images, 2,678 annotations.
Volume of dataset	Approx. 848 MB
Key index in dataset	Pest-induced hole bounding box coordinates (x_center, y_center, width, height), class ID = 0
Data accessibility	DOI:10.57760/sciencedb.29865; https://www.scidb.cn/detail?dataSetId=8954e40ec0eb4f3fb7dad889e982547f CSTR: 31253.11.sciencedb.29865; https://www.scidb.cn/s/RF7J3q
Financial support	None

Key words: Rehmannia glutinosa, Pest-induced Hole detection, image dataset, data annotation, YOLO, agricultural artificial intelligence

许琳娜, 黄婷, 郑丽萍, 费选. 地黄叶片虫害孔洞图像数据集（RPHD）的构建与基准评估[J]. 农业大数据学报, 2026, 8(2): 258-265.

XU LinNa, HUANG Ting, ZHENG LiPing, FEI Xuan. Construction and Benchmark Evaluation of the Rehmannia Leaf Pest-induced Hole Image Dataset (RPHD)[J]. Journal of Agricultural Big Data, 2026, 8(2): 258-265.

图/表 6

图1

图2

图3

表1

表2

表3

参考文献 25

[1]	杨玲, 王震, 向臻, 等. 豫西南地黄轮纹病的发生与防治措施. 河南农业, 2016(10): 35-36.
	YANG L, WANG Z, XIANG Z, et al. Occurrence and control measures of Rehmannia glutinosa ring spot disease in Southwest Henan. Henan Agriculture, 2016(10): 35-36.
[2]	杨洁, 赵荣兵, 宋爱青, 等. 地黄根腐病的发生规律与防治策略研究. 种子科技, 2025, 43(18): 177-179.
	YANG J, ZHAO R B, SONG A Q, et al. Study on occurrence regularity and control strategy of Rehmannia glutinosa root rot. Seed Science & Technology, 2025, 43(18): 177-179.
[3]	秦艳红, 文艺, 高素霞, 等. 河南省地黄病毒病病原鉴定及主要病毒的分子变异分析. 植物病理学报, 2024, 54(2): 469-475. doi: 10.13926/j.cnki.apps.001323
	QIN Y H, WEN Y, GAO S X, et al. Identification of the pathogens of Rehmannia glutinosa viral disease and molecular variation analysis of the major viruses. Acta Phytopathologica Sinica, 2024, 54(2): 469-475.
[4]	王湾湾. 基于深度学习技术的农业图像处理研究综述. 科技资讯, 2024, 22(16): 168-170.
	WANG W W. Review of research on agricultural image processing based on deep learning technology. Science & Technology Information, 2024, 22(16): 168-170.
[5]	郑远攀, 李广阳, 李晔. 深度学习在图像识别中的应用研究综述. 计算机工程与应用, 2019, 55(12): 20-36. doi: 10.3778/j.issn.1002-8331.1903-0031
	ZHENG Y P, LI G Y, LI Y. Survey of application of deep learning in image recognition. Computer Engineering and Applications, 2019, 55(12): 20-36. doi: 10.3778/j.issn.1002-8331.1903-0031
[6]	于俊伟, 翟付品. 河南工业大学储粮害虫图像数据集. 农业大数据学报, 2023, 5(2): 85-90. doi: 10.19788/j.issn.2096-6369.230213
	YU J W, ZHAI F P. Grain pest image dataset of Henan University of Technology. Journal of Agricultural Big Data, 2023, 5(2): 85-90.
[7]	李东亚, 王震鲁, 戴硕, 等. 棉花病虫害知识图谱构建数据集. 中国科学数据, 2025, 10(3): 1-10.
	LI D Y, WANG Z L, DAI S, et al. A dataset for constructing knowledge graph of cotton diseases and pests. China Scientific Data, 2025, 10(3): 1-10.
[8]	陈磊, 刘立波, 王晓丽. 2020年宁夏枸杞虫害图文跨模态检索数据集[J/OL]. 中国科学数据, 2022, 7(3). DOI: 10.11922/11-6035.nasdc.2021.0058.zh.
	CHEN L, LIU L B, WANG X L. A dataset of image-text cross-modal retrieval of Lycium barbarum pests in Ningxia in 2020[J/OL]. China Scientific Data, 2022, 7(3). DOI: 10.11922/11-6035.nasdc.2021.0058.zh.
[9]	管博伦, 张立平, 朱静波, 等. 农业病虫害图像数据集构建关键问题及评价方法综述. 智慧农业(中英文), 2023, 5(3): 17-34.
	GUAN B L, ZHANG L P, ZHU J B, et al. The key issues and evaluation methods for constructing agricultural pest and disease image datasets: A review. Smart Agriculture, 2023, 5(3): 17-34. doi: 10.12133/j.smartag.SA202306012
[10]	曹亮, 肖伟, 李湘丽. YOLO算法及其在农作物识别及病虫害检测应用综述. 仲恺农业工程学院学报, 2025, 38(6): 60-71.
	CAO L, XIAO W, LI X L. Review on YOLO algorithm and its application in crop identification and pest detection. Journal of Zhongkai University of Agriculture and Engineering, 2025, 38(6): 60-71.
[11]	万应霞, 燕振刚. 基于YOLO算法的农作物病虫害识别研究综述. 热带农业工程, 2024, 48(1): 25-28.
	WAN Y X, YAN Z G. Research review of crop diseases and pests identification based on YOLO algorithm. Tropical Agricultural Engineering, 2024, 48(1): 25-28.
[12]	蒋心璐, 陈天恩, 王聪, 等. 大田环境下的农业害虫图像小目标检测算法. 计算机工程, 2024, 50(1): 232-241. doi: 10.19678/j.issn.1000-3428.0067030
	JIANG X L, CHEN T E, WANG C, et al. Small object detection algorithm for agricultural pest images in field environments. Computer Engineering, 2024, 50(1): 232-241. doi: 10.19678/j.issn.1000-3428.0067030
[13]	邓镛. 新技术背景下无人机遥感技术在农业病虫害监测中的应用探讨. 四川农业与农机, 2024(5): 21-23.
	DENG Y. Application of UAV remote sensing technology in agricultural pest and disease monitoring under the background of new technology. Sichuan Agriculture and Agricultural Machinery, 2024(5): 21-23.
[14]	TOMASI C, MANDUCHI R. Bilateral filtering for gray and color images// Sixth International Conference on Computer Vision (ICCV). Bombay: IEEE, 1998: 839-846.
[15]	JIANG K, ZHANG B, ZHANG B, et al. An adaptive weight fusion low-light image enhancement based on HSV space[J/OL]. Multimedia Systems, 2025. DOI: 10.1007/s00530-025-01935-x.
[16]	张哲, 李佩霏, 吕宪勇, 等. 面向智慧农业的智能图像标注系统. 农业装备与车辆工程, 2024, 62(7): 145-152.
	ZHANG Z, LI P F, LYU X Y, et al. Intelligent image annotation system for smart agriculture. Agricultural Equipment & Vehicle Engineering, 2024, 62(7): 145-152.
[17]	马俊红, 刘冬梅, 李永亮, 等. 烟草病虫药害智能识别基准数据集构建及三维注意力模型设计. 中国烟草学报, 2021, 27(5): 52-60.
	MA J H, LIU D M, LI Y L, et al. Construction of benchmark data set for intelligent identification of tobacco pests, diseases, phytotoxicity and design of three dimensional attention model. Acta Tabacaria Sinica, 2021, 27(5): 52-60.
[18]	WANG A, CHEN H, LIU L, et al. YOLOv10:Real-time end-to- end object detection[J/OL]. arXiv, 2024. DOI: 10.48550/arXiv.2405.14458.
[19]	KHANAM R, HUSSAIN M. YOLOv11:An overview of the key architectural enhancements[J/OL]. arXiv, 2024. DOI: 10.48550/arXiv.2410.17725.
[20]	TIAN Y, YANG Q, YAO Y, et al. YOLOv12:Attention-centric real-time object detectors[J/OL]. arXiv, 2025. DOI: 10.48550/arXiv.2502.12524.
[21]	王宁, 智敏. 深度学习下的单阶段通用目标检测算法研究综述. 计算机科学与探索, 2025, 19(5): 1115-1140. doi: 10.3778/j.issn.1673-9418.2411032
	WANG N, ZHI M. Review of one-stage universal object detection algorithms in deep learning. Journal of Frontiers of Computer Science and Technology, 2025, 19(5): 1115-1140.
[22]	钟志峰, 彭宅琨, 黄培沛, 等. 多尺度特征感知的无人机小目标检测算法[J/OL]. 计算机工程, 2026. DOI: 10.19678/j.issn.1000-3428.0252593.
	ZHONG Z F, PENG Z K, HUANG P P, et al. Multi-scale feature-aware small object detection for UAV imagery[J/OL]. Computer Engineering, 2026. DOI: 10.19678/j.issn.1000-3428.0252593.
[23]	毛昕蓉, 徐霄. 基于改进YOLOv8s的航拍图像小目标检测算法. 微电子学与计算机, 2026, 43(3): 75-87.
	MAO X R, XU X. Small target detection algorithm for aerial images based on improved YOLOv8s. Microelectronics & Computer, 2026, 43(3): 75-87.
[24]	尹哲, 王广龙, 林森, 等. 基于改进YOLOv11的监控视频小目标检测算法. 沈阳理工大学学报, 2026, 45(2): 62-68.
	YIN Z, WANG G L, LIN S, et al. Small target detection algorithm in surveillance video based on improved YOLOv11. Journal of Shenyang Ligong University, 2026, 45(2): 62-68.
[25]	肖宇. 基于轻量卷积神经网络的农作物病虫害图像识别研究[D]. 鞍山: 辽宁科技大学, 2025.
	XIAO Y. Research on image recognition of crop diseases and insect pests based on lightweight convolutional neural network[D]. Anshan: University of Science and Technology Liaoning, 2025.

统计维度	原始版本	优化版本
图像数量	298	291
标注总数	5 059	2 678
平均每图标注数	16.98	9.20
图像平均分辨率	原始尺度	1024×1024
平均目标宽度	0.0158	0.0492
平均目标高度	0.0229	0.0488
平均目标占图比 (%)	0.0558	0.3867
有效图像区域占比 (%)	99.36	100.57
标注框长宽比标准差	0.6988	0.4227

类别	配置
操作系统	Windows 11
深度学习框架	PyTorch 2.5.0
编程语言	Python 3.11
CUDA与cuDNN	CUDA 11.8, cuDNN 9.1.0
CPU	13th Gen Intel(R) Core(TM) i7-13700H (2.40 GHz)
GPU	NVIDIA GeForce RTX 4060 Laptop GPU
内存	16 GB

数据集版本	模型	mAP@0.5	mAP@0.5-0.95	Precision	Recall
YOLOv10n	原始集	40.3%	19.4%	51.6%	41.6%
YOLOv10n	优化集	87.2%	57.1%	89.7%	75.0%
YOLOv11n	原始集	48.9%	22.3%	56.5%	39.3%
YOLOv11n	优化集	92.1%	62.7%	89.0%	85.4%
YOLO12n	原始集	51.2%	22.9%	58.5%	41.8%
YOLO12n	优化集	93.0%	62.2%	88.3%	88.3%