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 Crop Diseases, Pests and Weeds Based on Deep Learning is an agricultural informatization software product independently developed by Nanjing Green Technology Research Institute Co., Ltd. Leveraging cutting-edge artificial intelligence (AI) technologies including deep learning, image recognition and knowledge graphs, this product is designed to provide users with intelligent and precise diagnosis and prevention decision-making services for agricultural crop diseases, pests and weeds amid the emerging trend of mobile internet development. We have collected 3,289,078 original images of 700 disease types affecting 48 crop varieties; 1 million original images of 102 pest species; and over 3 million original images of more than 400 weed species. After undergoing meticulous filtering and processing, these original images meet the training requirements. We have also collected images of 35 disease types, 102 pest species and over 400 weed species. Following rigorous filtering and processing, these datasets also meet the training standards. By applying relevant image recognition algorithms for data augmentation, deep learning and model training, and relying on the crop diseases, pests and weeds image database, we have trained a model with the capability to identify crop diseases, pests and weeds using large-scale image recognition models. This enables agricultural producers to instantly identify the specific diseases, pests or weeds affecting their crops by taking photos with mobile phones, and obtain convenient and accurate information including details of the affected issues, prevention and control measures, medication and fertilization guidance and other relevant support.

搜集汇总

背景与挑战

背景概述

该数据集包含48种农作物700种病害的3289078张图片，102种虫害的100万张图片，以及400多种草害的300多万张图片，经过处理后可用于训练图像识别模型，支持农业生产者通过拍照识别病虫草害并提供防治建议。