Plant disease diagnosis system based on disease spot correlation

Abstract

The invention provides a disease spot correlation-based plant disease diagnosis system. The system comprises the following steps of S1, acquiring a picture through camera equipment, and uploading thepicture to a cloud server through a network; S2, for the picture obtained in the step S1, performing scaling and normalization preprocessing on the picture, and detecting a scab area existing in the uploaded picture by using a CenterNet target detection algorithm; S3, according to all the lesion areas obtained in the step S2, sequentially sending the lesion areas to a clustering feature extractionmodel to obtain a feature set; and S4, according to the feature set obtained in the step S3, calculating cosine similarity between each feature in the feature set and all features in the database, performing area weighted summation on all regions, and taking the category of the feature with the maximum weighted cosine similarity as a diagnosis result. The invention belongs to the technical fieldof computer vision, replaces a traditional identification mode directly based on a whole picture and realizes high-precision plant disease and insect pest identification.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a plant disease diagnosis system based on lesion correlation. Background technique [0002] There are many vegetations in our country, and agriculture is developed. Agriculture is an important pillar industry. For a long time, due to the harm of diseases and insect pests, it has caused huge economic losses and seriously affected the sustainable development of regional agricultural economy. The pests and diseases suffered by vegetation and crops are the main natural disasters in our country. Diagnosis and treatment of pests and diseases will not only help to increase crop yields and reduce economic losses, but also increase the vegetation coverage in our country. However, there are so many types of plant diseases and insect pests that it is difficult for ordinary people to make an accurate diagnosis of the diseased plants only through the diseased spots and insects on the ...

AI Technical Summary

Problems solved by technology

Since different plants may suffer from the same pest, there is a correlation between the types of pests and diseases among different plants. However, previous pest recognition algorithms directly diagnose the types of pests and diseases by directly identifying the entire picture, and do not use This correlation is used to improve the accuracy of the algorithm for identifying the types of pests and diseases. Therefore, when encountering plants suffering from pests and diseases that have never been seen before, the model needs to be retrained, which will greatly increase the training cost of the model and the generalization is poor.

Images