农作物病虫草害图像识别诊断大数据集

基于深度学习的农业病虫草害图像识别智能诊断大数据集是一款由南京绿色科技研究院有限公司自主研发的农业信息化软件产品。产品基于前沿的人工智能（AI）技术，包括深度学习、图像识别、知识图谱等，旨在为用户提供移动互联新兴形势下农业病虫害的智能精准诊断及防治决策服务。我们收集了48种农作物700种病害原始样本3289078张图片；102种虫害图片包括了原始样本100万张图片；400多种虫害图片包括了300多万张原始样本图片。这些原始图片经过细致的过滤和处理之后能够达到训练的要求。我们收集了35种病害图片，102种虫害图片，400多种草害图片。这些原始图片经过细致的过滤和处理之后能够达到训练的要求，利用相关图像识别算法进行数据增强、深度学习、训练等，基于农作物病虫草害图片数据库，使用图像识别大模型训练出具有识别病虫草害能力的模型，让农业生产者通过手机拍照就能立即获取农作物得了什么病虫草害，同时获得更加方便精准的了病虫草害信息、防治措施、防治办法、用药施肥等。

The large-scale dataset for intelligent diagnosis of agricultural diseases, pests and weeds based on image recognition is an agricultural informatization software product independently developed by Nanjing Green Technology Research Institute Co., Ltd. This product is based on cutting-edge artificial intelligence (AI) technologies including deep learning, image recognition and knowledge graphs, aiming to provide users with intelligent and accurate diagnosis and prevention decision-making services for agricultural diseases, pests and weeds under the new trend of mobile internet. We have collected 3,289,078 original sample images for 700 types of diseases affecting 48 crop varieties; 1,000,000 original sample images for 102 types of pests; and more than 3,000,000 original sample images for over 400 types of pests. These original images have undergone meticulous filtering and processing to meet the requirements of model training. Additionally, we have collected images for 35 types of crop diseases, 102 types of pests, and over 400 types of weed pests, all of which have been thoroughly filtered and processed to meet training standards. Using relevant image recognition algorithms for data augmentation, deep learning and model training, and based on the image database of crop diseases, pests and weeds, we have trained a model capable of identifying agricultural diseases, pests and weeds via large-scale image recognition models. This allows agricultural producers to immediately obtain information about the specific diseases, pests or weeds affecting their crops by taking photos with mobile phones, as well as convenient and accurate information related to the pests and diseases, prevention and control measures, medication and fertilization guidance.

搜集汇总

背景与挑战

背景概述

该数据集是一个基于深度学习的农业病虫草害图像识别智能诊断大数据集，由南京绿色科技研究院有限公司开发，涵盖48种农作物的700种病害、102种虫害和400多种草害，总计原始样本图片数量超过700万张。它利用前沿人工智能技术，包括图像识别和知识图谱，通过训练模型实现智能精准诊断，帮助农业生产者通过手机拍照快速获取病虫草害信息及防治措施。数据集经过细致过滤和处理，适用于图像识别算法的训练，旨在提升农业病虫害防治的效率和准确性。