农业大数据学报 ›› 2022, Vol. 4 ›› Issue (1): 114-118.doi: 10.19788/j.issn.2096-6369.220117

• 数据论文 • 上一篇    

桔小实蝇等六种常见果园害虫图像数据集

张翔鹤1,2(), 王晓丽1,2,3, 刘婷婷1,2,3, 胡林1,2, 樊景超1,2,3()   

  1. 1.中国农业科学院农业信息研究所,北京 100081
    2.国家农业科学数据中心,北京 100081
    3.农业农村部农业大数据重点实验室北京 100081
  • 收稿日期:2021-12-20 出版日期:2022-03-26 发布日期:2022-06-29
  • 通讯作者: 樊景超 E-mail:zhxianghe@163.com;fanjingchao@caas.cn
  • 作者简介:[1] 张翔鹤|张翔鹤,女,硕士,研究生,研究方向:农业科学数据管理; E-mail: zhxianghe@163.com|王晓丽,刘婷婷,等. 桔小实蝇等六种常见果园害虫图像数据集[DB/OL].国家农业科学数据中心.DOI:10.12205/asda.j00003.00008.|王晓丽,刘婷婷,等. 桔小实蝇等六种常见果园害虫图像数据集[DB/OL].国家农业科学数据中心.DOI:10.12205/asda.j00003.00008.
  • 基金资助:
    中国农业科学院创新工程:数据整合与应用服务研究(2020CX017)

Image Data Set of Six Common Orchard Pests such as Bactrocera Dorsalis

Xianghe Zhang1,2(), Xiaoli Wang1,2,3, Tingting Liu1,2,3, Lin Hu1,2, Jingchao Fan1,2,3()   

  1. 1.Agricultural Information Institute of Chinese Academy of Agricultural Sciences, Beijing 100081
    2.National Agriculture Science Data Center, Beijing 100081
    3.Key Laboratory of Big Agri-Data, Ministry of Agriculture, Beijing 100081
  • Received:2021-12-20 Online:2022-03-26 Published:2022-06-29
  • Contact: Jingchao Fan E-mail:zhxianghe@163.com;fanjingchao@caas.cn

摘要:

使用机器视觉方法进行虫害识别是果园害虫防控或治理的必然需求。目前对果园害虫图像数据的采集,多数品种单一,分辨率参差不齐。并且仅收集害虫原始图像数据,同时包含原始图像和机器识别显著图图像的数据集极少。本数据集包括桔小实蝇、金龟子、梨小食心虫、青叶蝉、星天牛和柑桔大实蝇六种常见害虫的图像数据,共计2412张。其中原始图像1613张,未经处理。经过反卷积方法处理的图像,剔除特征不显著的图像后,保留特征显著图像,共计799张。该数据集可为果园害虫的识别分类研究提供数据基础。

关键词: 果园, 害虫识别, 图像数据, 机器识别

Abstract:

It is essential to use machine vision method for pest identification in orchard pest control and management. At present, most of the orchard pest image data collection centre on a single type and the resolution is inconsistent. In addition, only the original image data of pests are collected, and few data sets contain both the original image and the salient image of machine recognition. This data set includes 2412 image data of six common pests, such as bactrocera dorsalis, chafer, grapholitha molesta, leaf hopper, long icorn and bactrocera minax. Among them, 1613 original images were unprocessed. For images processed by deconvolution method, a total of 799 images with significant features were retained after eliminating the images with insignificant features. In conclusion, the data set can provide a data basis for the identification and classification of orchard pests.

Key words: orchard, pests identification, image data, machine recognition

中图分类号: 

  • S436.6