Potato disease detection method and system based on image recognition

Abstract

The invention relates to a potato disease detection method based on image recognition, which comprises the following steps that: step 1, a client acquires pictures of potato leaves, rhizomes and potato blocks to be recognized, and uploads the pictures to a server; step 2, the server preprocesses the picture, and then sends the preprocessed picture to a deep learning model; step 3, the deep learning model extracts high-dimensional features from the preprocessed image and then identifies and calibrates the types and positions of the potato scab contained in the image; And step 4, the identifiedscab position and the category thereof on the picture are superposed as marks to form a detection result picture for downloading and displaying by the client.

Description

technical field [0001] The invention belongs to the fields of computer vision and pattern recognition, and in particular relates to the technology of detecting and locating objects of interest based on deep learning, and more specifically, relates to a method and system for detecting potato diseases based on image recognition. Background technique [0002] With the advent of computer vision and dramatic improvements in the accuracy and speed of deep learning networks, more image recognition tasks in everyday life can use these techniques to meet engineering needs. The same is true for potato disease detection. In order to allow users who lack relevant knowledge to quickly identify the type and development degree of potato diseases, applying deep learning models for image recognition, especially image detection, is an important development trend of disease detection. [0003] How to let farmers know the professional disease detection results and treatment plan within 10 secon...

AI Technical Summary

Problems solved by technology

[0012] Fifth: This solution requires an additional download of the self-developed APP, which will inevitably lead to incompatibility between different mobile phone systems and versions;

[0016] First: Only potato leaves can be identified, but roots and tubers cannot be identified;

[0017] Second: Like the scheme 1, it is also realized by the idea of ​​image classification, and there are problems similar to scheme 1, such as the inability to accurately locate;

[0018] Second: The accuracy of the traditional SVM classifier is much lower than that based on deep learning;

[0019] Third: The SVM algorithm is difficult to implement for large data training samples. In practical applications, the image data of potato blight leaves is very large, and the deep learning model can make full use of these data;

[0020] Second: The SVM algorithm has difficulties in solving multi-classification problems, and this scheme does not apply related technologies such as decision trees to overcome this defect

In the potato disease detection problem, not only common diseases such as early blight and late blight are involved, but if other viral diseases that are difficult to distinguish in appearance characteristics are added, its effect will be far worse than that of deep learning models that are good at handling multi-category tasks

Images