农业大数据学报 ›› 2024, Vol. 6 ›› Issue (4): 558-563.doi: 10.19788/j.issn.2096-6369.100026

• • 上一篇    下一篇

2022年黑龙江小麦、玉米、水稻苗期图像数据集

秦佳乐1,2(), 苑江浩3, 宋国柱1, 姚鸿勋4, 郭雷风2,*(), 王晓丽2,5,6,*()   

  1. 1.山西农业大学软件学院,山西太谷 030801
    2.中国农业科学院农业信息学研究所,北京 100081
    3.国家粮食和物资储备局科学研究院,北京 100037
    4.哈尔滨工业大学计算机科学与技术学院,哈尔滨 150001
    5.国家农业科学数据中心,北京 100081
    6.三亚中国农业科学院国家南繁研究院,海南三亚 572024
  • 收稿日期:2024-03-26 接受日期:2024-04-25 出版日期:2024-12-26 发布日期:2024-12-02
  • 通讯作者: 郭雷风,E-mail: guoleifeng@caas.cn
    王晓丽,E-mail: wangxiaoli@caas.cn
  • 作者简介:秦佳乐,E-mail:1104291012@qq.com
  • 基金资助:
    国家科技创新2030重大项目(2021ZD0110901);内蒙古自治区科技计划项目(2021GG0341)

Image Dataset of Wheat, Corn, and Rice Seedlings in Heilongjiang Province in 2022

QIN JiaLe1,2(), YUAN JiangHao3, SONG GuoZhu1, YAO HongXun4, GUO LeiFeng2,*(), WANG XiaoLi2,5,6,*()   

  1. 1. School of Software, Shanxi Agricultural University, Taigu 030801, Shanxi, China
    2. Institute of Agricultural Informatics, Beijing 100081, China
    3. Academy of National Food and Strategic Reserves Administration, Beijing, 100037, China
    4. Computer Science and Technology, Harbin Institute of Technology, Harbin 150001, China
    5. National Agriculture Science Data Center, Beijing 100081, China
    6. National Nanfan Research Institute (Sanya), Chinese Academy of Agricultural Sciences, Sanya 572024, Hainan, China
  • Received:2024-03-26 Accepted:2024-04-25 Published:2024-12-26 Online:2024-12-02

摘要:

大田作物大多采取露地种植方式。东北地区全年温度较低,在作物苗期如果出现日照和降雨量大范围的波动,则十分容易导致农作物出现苗弱苗小、根系长势弱以及发育不全和生长缓慢等现象。若能对农作物苗期实时监测和管理,及时掌握其生长状态及其环境情况,便可及早做出决策。本研究于2022年5月9日—2022年6月16日期间,对试验田内11个气象站的小麦、玉米和水稻、小麦苗期图像进行采集,通过整理和筛查后形成的数据量约为2.59 GB,其中可见光RGB 1.48 GB,近红外光谱 1.11 GB。本数据集可以通过RGB可见光数据和近红外光谱数据完成对作物的叶龄识别,将提取出的特征(颜色特征、图像特征、纹理特征、植被指数)带入机器学习回归模型中进行分析预测,同时本数据集还适用于构建作物识别或幼苗识别的卷积神经网络模型,以进一步精准实现作物检测及插秧后漏苗、补苗等研究。

数据摘要:

项目 描述
数据库(集)名称 2022年黑龙江小麦、玉米、水稻苗期图像数据集
所属学科 农业科学
研究主题 计算机视觉
数据时间范围 2022年5月—2022年7月
时间分辨率 1天
数据类型与技术格式 .jpg
数据库(集)组成 数据集由两部分数据组成,其一为大田作物可见光RGB图像数据集1.48 GB,其二是大田作物近红外光谱图像数据集1.11 GB,均包含38天的数据。
数据量 2.59 GB
主要数据指标 RGB图像和近红外光谱图像
数据可用性 CSTR: https://cstr.cn/17058.11.sciencedb.agriculture.00092
DOI: https://doi.org/10.57760/sciencedb.agriculture.00092
NASDC访问链接: https://agri.scidb.cn/ ,限制性获取
经费支持 国家科技创新2030重大项目(2021ZD0110901);内蒙古自治区科技计划项目(2021GG0341)

关键词: 黑龙江, 苗期数据集, 小麦, 玉米, 水稻, 可见光图像, 近红外光谱图像

Abstract:

During the cultivation process, most field crops are typically grown in open fields. The northeastern region of China experiences relatively low temperatures throughout the year. During the seedling stage of crops, significant fluctuations in sunlight and rainfall can easily lead to issues such as weak and stunted seedlings, poorly developed root systems, and slow growth. Timely monitoring and management of crops during the seedling stage can help in understanding their growth status and environmental conditions, enabling prompt decision-making.Experimental data was collected from May 9, 2022, to June 16, 2022. RGB cameras installed at 11 meteorological stations in the experimental fields collected data seven times a day at 6:00, 8:00, 10:00, 12:00, 14:00, 16:00, and 18:00. The images were captured at a height of 2.4 meters with a field of view angle of 90°, covering an area of 4.4 meters in length and 2.5 meters in width. Photography was mainly conducted through natural light conditions with a downward vertical perspective.After organizing and screening, the dataset comprises approximately 2.59 GB of data, including 1.48 GB of visible light RGB data and 1.11 GB of near-infrared spectral data. This dataset enables leaf age identification through visible light RGB data and near-infrared spectral data. Extracted features (color features, image features, texture features, vegetation indices) can be inputted into machine learning regression models for analysis and prediction. Moreover, this dataset is suitable for constructing convolutional neural network models for crop recognition or seedling identification, facilitating precise crop detection and further research on issues such as missed or replanted seedlings after transplanting.

Data summary:

Items Description
Dataset name Image Dataset of Wheat, Corn, and Rice Seedlings in Heilongjiang Province in 2022
Specific subject area Agricultural science
Research Topic Computer vision
Time range May 2022-July 2022
Temporal resolution 1 day
Data types and technical formats .jpg
Dataset structure The dataset consists of two parts of data, one is the field crop visible light RGB image data set, and the other is the field crop multispectral near-infrared image data set, of which: 1. The field crop image data contains data within 38 days, and the data volume is 1.48G; 2. Daejeon near-infrared spectral data within 38 days, the data volume is 1.11G.
Volume of dataset 2.59 GB
Key index in dataset RGB images and near-infrared spectral images
Data accessibility CSTR: https://cstr.cn/17058.11.sciencedb.agriculture.00092
DOI: https://doi.org/10.57760/sciencedb.agriculture.00092
hNASDC Access link: https://agri.scidb.cn/, restricted access
Financial support National Key R&D Program of China (2021ZD0110901); Science and Technology Planning Project of Inner Mongolia Autonomous Region (2021GG0341)

Key words: Heilongjiang, seedling stage dataset, wheat, corn, rice, visible light image, near-infrared spectral image