Crop fruit late blight identification method based on computer vision

Abstract

The invention relates to a crop fruit late blight identification method based on computer vision. The method comprises the following steps: collecting fruit plant images; acquiring a fruit detection model, including labeling the collected image to generate a data set taking the fruit as a detection target, and constructing and training an image detection model; obtaining a fruit late blight identification model, including preparation of a fruit late blight identification data set, establishment of a fruit late blight classification network and model training; and identifying the fruit with late blight and outputting the proportion of the fruit in the total detected fruit. According to the method provided by the invention, disease judgment of plants in a visible range is realized, and the positions of diseased fruits are accurately positioned. The problem of unbalanced proportion of diseased fruits is solved through a multi-model fusion method, the detection accuracy of crop fruits is improved, the ratio of the marked diseased fruits to the total detection fruit quantity is calculated, and a data analysis scheme can be provided for late late blight attack hazard assessment.

Description

technical field [0001] The invention relates to the technical field of crop disease identification, in particular to a method for identifying late blight of crop fruits based on computer vision. Background technique [0002] Crop late blight is one of the most common and extremely harmful diseases in the process of plant growth, which seriously affects the yield and quality of crop fruits. Therefore, it is particularly important to effectively identify crop late blight and detect fruits suffering from crop late blight. [0003] In the prior art, the detection of crop late blight mainly relies on the identification of agricultural professionals. However, due to the lack of professional agronomic knowledge, ordinary farmers cannot accurately identify whether the crops are suffering from late blight, which delays the disease, resulting in a decline in crop fruit quality and yield; hiring professional agronomy experts will increase additional economic burdens for ordinary farme...

AI Technical Summary

Problems solved by technology

In recent years, there have been more and more studies on the identification of crop fruit late blight disease, but most of them are based on a single deep learning network

There are the following defects: the amount of data is small, which brings great difficulties to model optimization; affected by the unbalanced amount of crop fruit data, the collected images generally have more normal fruits, but fewer fruits with late blight lesions, single The image detection model has a high fruit detection rate, but a low fruit classification recognition rate

Images