Rice disease image detection method integrating multiple context deep learning models

Abstract

The invention relates to a rice disease image detection method integrating multiple context deep learning models. Compared with the prior art, the rice disease image detection method can assist in eliminating a defect that rice disease image detection does not consider relevant condition factors of the rice disease image detection. The method of the invention includes the following steps that: training samples are collected and pre-processed; rice image context information is integrated on the basis of deep learning, so that a rice disease image detection model is obtained; the space, location, temperature and humidity information of a rice image to be processed is collected and preprocessed; and the specific position of a disease in the rice image is marked. According to the rice diseaseimage detection method integrating multiple contextual deep learning models of the invention, the features of the disease image itself are considered, and relevant factors such as space, time, temperature and humidity at the time of the collection of the disease image are also considered; a rice disease image global context training model, a rice disease image local context training model and a rice disease attribute constrained context training model are integrated; and therefore, the detection and recognition capacity of the method of the invention for the rice disease image under a field complex application condition can be improved, and the detection rate of the rice disease image can be improved.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a rice disease image detection method that integrates multiple context deep learning models. Background technique [0002] How to accurately detect and identify rice diseases has always been a problem that plagues crop forecasting. Due to the different locations and degrees of rice diseases in the farmland environment, the difficulty of manual visual inspection has been increased. At present, the detection of rice diseases is mainly done by a small number of plant protection experts and agricultural technicians. However, the rice image background in the farmland environment is complex, and the agricultural technicians are affected by knowledge level factors, so it is difficult to guarantee the accuracy with the naked eye. At the same time, due to the complex image background of rice disease in the natural environment, coupled with the influence of light and shadow, the ...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the defect that rice disease detection in the prior art does not consider its related conditional factors, and provide a rice disease image detection method that integrates multiple context deep learning models to solve the above problems

Images