首页 > 分享 > 基于自然背景的蚜虫图像数据集

基于自然背景的蚜虫图像数据集

花匠小妙招
2025-04-05 12:52

摘要：

蚜虫的发生是影响农作物产量和质量的重要原因之一。对蚜虫进行检测和计数是对虫害早发现、早治理的重要环节。随着信息技术的发展，已经有专家学者利用计算机视觉感知技术对农业害虫进行识别研究，并取得了一定的进展。高质量、大规模的基础数据对计算机视觉的发展往往能够起到决定性作用，缺少高质量、大规模的基础图像数据是蚜虫精准识别研究面临的难题。蚜虫是一类重要的农业害虫，具有尺寸微小、密集分布、虫间遮挡和同种多形态等特征，这些特征对于蚜虫的检测与计数又是一项严峻的挑战。本文提供了包括桃粉蚜、桃蚜、棉蚜、禾谷缢管蚜等13种农业蚜虫数据集，共6287张高清原始图像。这些蚜虫图像是利用单反相机在自然大田环境中采集、以文件夹形式进行存储、经过从事图像数据管理的专业人员清洗和整理、并由植保专家对其进行鉴定和分类的，保障了数据的高质量和可靠性。该数据集可为蚜虫的识别、检测计数和分类提供数据基础。

数据摘要：

项目描述数据库（集）名称基于自然背景的蚜虫图像数据集(Aphis 13) 所属学科植物保护研究主题蚜虫数据时间范围 2013—2023年数据地理空间覆盖中国境内数据类型与技术格式数据类型：图像；技术格式：*.jpg 数据库（集）组成数据集包括桃粉蚜、桃蚜、棉蚜、禾谷缢管蚜、绣线菊蚜、花生蚜、莴苣指管蚜、荻草谷网蚜、甘蓝蚜、萝卜蚜、玉米蚜、核桃全斑蚜和梨大绿蚜等13类蚜虫图像，共6287张数据量 16.8 GB 数据可用性 CSTR: https://cstr.cn/17058.11.sciencedb.agriculture.00030
DOI: https://doi.org/10.57760/sciencedb.agriculture.00030 经费支持国家自然科学基金面上项目“知识迁移与因果推理启发的小样本害虫图像识别研究”（项目编号：32171888）；安徽省农业科学院科研计划项目“农业智能化技术研发中心”(2023YL014)

关键词: 蚜虫, 计算机视觉, 图像数据

Abstract:

Agricultural pests are important reasons affecting crop yield and quality. Aphid is an important group of agricultural pest. Detecting and counting aphids is an important link for early detection and management of this pest. With the development of information technology, many experts and scholars have conducted extensive research on the identification of agricultural pests using computer vision, and have made certain progress. High-quality and large-scale basic data often play a decisive role in the development of computer vision, but the lack of this kind of image data is one of the challenges faced by pest identification. Aphids have features such as small size, dense distribution, inter insect shelter, and multiple forms of same species. These features also pose a serious challenge for the detection and counting of aphids. This article provides a total of 6287 high-definition original images, including a dataset of 13 agricultural pests (aphids) including peach aphid, cotton aphid, and grain constrictor aphid, etc. These aphid images were collected using DSLR cameras in a natural field environment. In order to ensure the high quality and reliability of the data, these images are cleaned and organized by professional personnel, and identified and classified by experts in the field of plant protection. This dataset can provide a data foundation for recognition, detection, counting and classification of aphids.

Data summary:

Items Description Dataset name Aphid Image Dataset Based on Natural Background Specific subject area Plant protection Research topic Aphid Time range 2013-2023 Geographical scope China Data types and technical formats Data type: image; Technical formats:*.jpg Dataset structure The dataset contains a total of 6287 images of 13 types of aphids, including Hyalopterus amygdali, Myzus persicae, Aphis gossypii, Rhopalosiphum padi, Aphis spiraecola, Aphis craccivora, Uroleucon formosanum, Sitobion miscanthi, Brevicoryne brassicae, Lipaphis erysimi, Rhopalosiphum maidis, Panaphis juglandis, and Nippolachnus piri. Volume of data 16.8 GB Data accessibility CSTR: https://cstr.cn/17058.11.sciencedb.agriculture.00030
DOI: https://doi.org/10.57760/sciencedb.agriculture.00030 Financial support General Program of National Natural Science Foundation of China “Research on Few-shot Pest Recognition Inspired by Knowledge Transfer and Causal Reasoning”（32171888）
Anhui Academy of Agricultural Sciences Research Platform Project “Agricultural Intelligent Technology Research and Development Center”（2023YL1014）

Key words: aphid, computer vision, image data