Data identification system based on disease identification model

Abstract

The invention discloses a data identification system based on a disease identification model, and particularly relates to the field of data identification, and the system comprises an image data collection module, a cucumber leaf image data processing module, a cucumber leaf image data analysis and calculation module, a feature extraction module, a disease identification model establishment module, a system integration module and a control end display module. According to the method, cucumber leaf images are collected through a digital camera, feature values are extracted after processing, analysis and calculation, and different diseases are classified and recognized through a deep learning model; finally, the system is integrated into a computer image processing system, an identification result is displayed through a control end display module, and meanwhile disease information is sent to a control end.

Description

Technical Field

[0001] The present invention relates to the technical field of data recognition, and more specifically, to a data recognition system based on a disease recognition model. Background Art

[0002] Big data analysis technology is a technology used to process massive amounts of data. It analyzes and mines large and complex data sets to discover potential value and connections. Big data technology usually involves data collection, storage, processing, analysis and visualization, aiming to help users gain deeper insights, more accurate predictions and more effective decisions.

[0003] The traditional methods of disease identification and diagnosis are mainly the following: one is to identify through agricultural disease atlas, the second is to read agricultural disease books, and the third is to search through classification retrieval tables.

[0004] However, it still has some shortcomings in actual use. For example, the traditional method requires professionals to have a lot of professional knowledge, and when reading agricultural disease books, the types of diseases are not complete, the reading speed is slow, and it is impossible to find agricultural diseases in all aspects. Therefore, a data recognition system based on disease recognition model came into being. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a data recognition system based on a disease recognition model to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] Image data acquisition module: used to use a shooting tool to shoot cucumber leaf images under natural light conditions, and transmit the shot cucumber leaf images to the image data processing module;

[0008] Cucumber leaf image data processing module: used to receive the cucumber leaf image collected by the image data collection module for image processing, and transmit the processed image to the feature extraction module;

[0009] Cucumber leaf image data analysis and calculation module: used to receive the processed image in the cucumber leaf image data processing module, and analyze and calculate the processed image data;

[0010] Feature extraction module: used to receive the data and images analyzed and calculated in the cucumber leaf image data analysis and calculation module, and extract features of the data and images;

[0011] Disease recognition neural network model establishment and training module: used to establish a disease recognition model and train the model to classify and identify different diseases using feature-extracted data;

[0012] System integration module: used to integrate the trained cucumber leaf disease recognition model into the computer image processing system;

[0013] Control end display module: used to identify the disease type in the disease model and send the disease type to the control end.

[0014] Preferably, in the image data acquisition module, the shooting tool is a digital camera, which uses automatic exposure mode under natural lighting conditions in the field to capture images of leaf diseases of cucumbers during the growth period with a resolution of 3840x2160 pixels, and transmits the images to a computer in BMP format.

[0015] Preferably, in the cucumber leaf image data processing module, the collected image is subjected to image denoising, and the method of performing image denoising using Gaussian filtering is specifically as follows:

[0016] Determine the Gaussian kernel size: First determine the size of the Gaussian kernel, that is, the size of the Gaussian filter; the size of the Gaussian kernel is an odd number, such as 3x3, 5x5, etc.;

[0017] Calculate the Gaussian kernel: According to the formula of the Gaussian function, calculate the weight at each position in the Gaussian kernel.

[0018] Image and Gaussian kernel convolution: The calculated Gaussian kernel is convolved with the image; for each pixel in the image, the surrounding pixels are weighted averaged according to the weight of the Gaussian kernel to obtain the new value of the pixel;

[0019] Boundary processing: At the pixel points at the image boundary, it may not be possible to find enough neighboring pixels for calculation. In this case, special processing of the boundary is required, such as using mirror extension, boundary pixel duplication, etc.

[0020] Output processing: According to actual needs, the output image is further post-processed, such as image enhancement.

[0021] Preferably, the cucumber leaf image data analysis and calculation module performs grayscale processing on the processed image, converts it into a grayscale image, and then performs segmentation processing on the grayscale image.

[0022] Preferably, in the feature extraction module, the three component features in the HIS model are extracted, and the method for converting RGB to HIS is specifically as follows:

[0023] The calculation method of hue value is as follows:

[0024] Among them, H represents the hue value, R represents the red value in the RGB model, G represents the green value in the RGB model, and B represents the blue value in the RGB model;

[0025] The specific calculation method of brightness value is:

[0026] Among them, I represents the brightness value, R represents the red value in the RGB model, G represents the green value in the RGB model, and B represents the blue value in the RGB model;

[0027] The specific calculation method of saturation is:

[0028]

[0029] Value, G represents the green value in the RGB model, and B represents the blue value in the RGB model;

[0030] The calculation method of texture feature value is as follows:

[0031] Where W represents the texture eigenvalue, N(l, θ) represents the energy value in the texture eigenvalue, C(l, θ) represents the entropy value in the texture eigenvalue, and G(l, θ) represents the moment of inertia value in the texture eigenvalue;

[0032] The specific method for calculating the energy value in the texture feature is:

[0033] N(l,θ)=∑ i，j K(i, j|l, θ) 2 , where N(l, θ) represents the energy value in the texture feature value, K(i, j|l, θ) represents the element value, (i, j) represents the probability of the grayscale layer appearing, l represents the distance, and θ represents the direction angle;

[0034] The specific method for calculating the entropy value in texture features is:

[0035] C(l,θ)=∑ i，j K(i, j|l, θ)-logK(i, j|l, θ), where C(l, θ) represents the entropy value in the texture feature value, K(i, j|l, θ) represents the element value, (i, j) represents the probability of the grayscale layer appearing, l represents the distance, and θ represents the direction angle;

[0036] The calculation method of the moment of inertia value in the texture feature is as follows:

[0037] G(l,θ)=∑ i，j (ij) 2K(i, j|l, θ), where G(l, θ) represents the moment of inertia value in the texture feature value, K(i, j|l, θ) represents the element value, (i, j) represents the probability of occurrence of the grayscale layer, l represents the distance, and θ represents the direction angle.

[0038] Preferably, in the disease recognition neural network model establishment and training module, a disease recognition neural network model is established, and the disease value is calculated by:

[0039] Among them, Z represents the disease value, H represents the hue value, I represents the brightness value, S represents the saturation, W represents the texture feature value, N(l, θ) represents the energy value in the texture feature value, C(l, θ) represents the entropy value in the texture feature value, and G(l, θ) represents the moment of inertia value in the texture feature value;

[0040] The specific method of constructing a neural network model for disease identification is as follows:

[0041] Select the BP neural network structure, and use the disease value, color feature value and texture feature value as the input value of the neural network, denoted as v1, v2, v3;

[0042] Input the input pattern v to the input layer neurons, and forward calculate the values ​​of the hidden layer neurons. The calculation method of the values ​​of the hidden layer neurons is as follows:

[0043] Among them, E1 represents the value of the hidden layer neuron, U represents the number of neurons in the input layer, d represents the dth neuron in the input and output layer, and DI od It is represented by the connection weights from the input layer to the hidden layer neurons determined after training, FI o It is represented by the threshold value of the input layer to the hidden layer neurons determined after training;

[0044] The output value of the hidden layer neuron is forward calculated. The calculation method of the value of the output layer neuron is as follows:

[0045] Among them, P a Represented as the value of the output layer neuron, represents the number of neurons in the output layer, o represents the oth neuron in the hidden layer, DT od It is represented by the connection weights from the hidden layer neurons to the output layer neurons determined after training, FI d It is represented by the threshold from hidden layer neurons to output layer neurons determined after training;

[0046] The output layer neuron value P calculated by the BP network a , a=1, 2, ..., X2, compare their sizes, and take the final output as neuron P max, where P max =max(t a ), a=1, 2, ..., X2, and the corresponding image category is the final classification result, and the final result is transmitted to the control end display module.

[0047] Preferably, the system integration module includes the following units:

[0048] Model export and deployment unit: First, you need to export the trained recognition model and save the model's weight parameters and structure information as a file; common formats include TensorFlow's .pb file and PyTorch's .pth file; deploy the model to the server according to the selected deployment method of the target system;

[0049] Interface design and data format processing unit: Design the interface for model calls to ensure that the system can interact with the model; unify the format requirements of input data, including the size and format of the image, and the format of the output results. The format of the output results refers to the classification results and disease location information; it is also necessary to handle data preprocessing and postprocessing to ensure that the input data meets the requirements of the model, and explain and display the output results;

[0050] Integration test and optimization unit: After integrating the model into the computer image processing system, conduct system-level integration testing to verify the performance of the model in actual application scenarios; optimize and adjust according to the test results to ensure that the overall performance of the system meets expectations;

[0051] Continuous update and maintenance unit: With the accumulation of disease data and the training of models, the system regularly updates the models to improve recognition accuracy. At the same time, it promptly handles errors and problems that arise in actual applications of the models to ensure the stability and reliability of the system.

[0052] Preferably, in the control end display module, a receiving port is set on the control end for receiving disease type information transmitted from the disease identification model, and the control end is a computer terminal; the received disease type information is displayed on the control end, and the type information is displayed in the form of graphics and text, and the name, picture and corresponding diagnosis method of the disease type are displayed on the control end.

[0053] Technical effects and advantages of the present invention:

[0054] The present invention collects cucumber leaf images through a digital camera, extracts characteristic values ​​after processing and analytical calculation, and uses a deep learning model to classify and identify different diseases; finally, the system is integrated into a computer image processing system, and the recognition result is displayed through a control end display module, and the disease information is sent to the control end. Through the processing of the model and the computer, agricultural diseases can be found in all aspects, and the efficiency is improved, and the probability of human errors is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic diagram of system module connection of the present invention.

[0056] Figure 2 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] See also Figure 1 As shown, the present invention relates to a data recognition system based on a disease recognition model, which includes an image data acquisition module, a cucumber leaf image data processing module, a cucumber leaf image data analysis and calculation module, a feature extraction module, a disease recognition neural network model establishment and training module, a system integration module, and a control end display module.

[0059] The image data acquisition module is connected to the cucumber leaf image data processing module, the cucumber leaf image data processing module is connected to the cucumber leaf image data analysis and calculation module, the cucumber leaf image data analysis and calculation module is connected to the feature extraction module, the feature extraction module is connected to the disease identification neural network model establishment and training module, the disease identification neural network model establishment and training module is connected to the system integration module, and the system integration module is connected to the control end display module.

[0060] The image data acquisition module is used to use a shooting tool to shoot cucumber leaf images under natural light conditions, and transmit the shot cucumber leaf images to the image data processing module;

[0061] In the image data acquisition module, the shooting tool is a digital camera. Under natural lighting conditions in the field, the automatic exposure mode is used to capture images of leaf diseases of cucumbers during the growth period with a resolution of 3840x2160 pixels, and the images are transmitted to the computer in BMP format.

[0062] The cucumber leaf image data processing module is used to receive the cucumber leaf image collected by the image data collection module for image processing, and transmit the processed image to the feature extraction module;

[0063] In the cucumber leaf image data processing module, the collected image is subjected to image denoising, and the method of using Gaussian filtering to perform image denoising is specifically as follows:

[0064] Determine the Gaussian kernel size: First determine the size of the Gaussian kernel, that is, the size of the Gaussian filter; the size of the Gaussian kernel is an odd number, such as 3x3, 5x5, etc.;

[0065] Calculate the Gaussian kernel: According to the formula of the Gaussian function, calculate the weight at each position in the Gaussian kernel.

[0066] Image and Gaussian kernel convolution: The calculated Gaussian kernel is convolved with the image; for each pixel in the image, the surrounding pixels are weighted averaged according to the weight of the Gaussian kernel to obtain the new value of the pixel;

[0067] Boundary processing: At the pixel points at the image boundary, it may not be possible to find enough neighboring pixels for calculation. In this case, special processing of the boundary is required, such as using mirror extension, boundary pixel duplication, etc.

[0068] Output processing: According to actual needs, the output image is further post-processed, such as image enhancement.

[0069] The cucumber leaf image data analysis and calculation module is used to receive the processed image in the cucumber leaf image data processing module, and analyze and calculate the processed image data;

[0070] The cucumber leaf image data analysis and calculation module performs grayscale processing on the processed image, converts it into a grayscale image, and then performs segmentation processing on the grayscale image. The specific method for processing the grayscale image is:

[0071] The grayscale image of size M×N is divided into two areas: unhealthy and healthy. The range of the image segmentation threshold is (1, 2, 3...B-1). The calculation method of the image segmentation threshold is as follows:

[0072] Y0=α1(h1-h0)+α2(h2-h0), where Y0 represents the image segmentation threshold, α1 represents the proportion of the unhealthy area in the image, α2 represents the proportion of the healthy area in the image, h1 represents the average grayscale of the unhealthy area, h2 represents the average grayscale of the healthy area, and h0 represents the average grayscale of the entire image;

[0073] Among them, the calculation method of the average grayscale of the entire image is specifically as follows:

[0074] Among them, h0 represents the average grayscale of the entire image, B(m) represents the proportion of grayscale m in the image, and m represents the grayscale;

[0075] The specific method for calculating the proportion of the unhealthy area in the image is:

[0076] Among them, α1 represents the proportion of unhealthy areas in the image, B(m) represents the proportion of gray level m in the image, m represents the gray level, and b represents the image segmentation into b categories;

[0077] The specific method for calculating the proportion of the healthy area in the image is:

[0078] Among them, α2 represents the proportion of healthy area in the image, B(m) represents the proportion of gray level m in the image, m represents the gray level, and b represents the image segmentation into b categories;

[0079] The calculation method of the average grayscale of the unhealthy area is as follows:

[0080] Among them, h1 represents the average grayscale of the unhealthy area, α1 represents the proportion of the unhealthy area in the image, B(m) represents the proportion of grayscale m in the image, m represents the grayscale, and b represents the image segmentation into b categories;

[0081] The calculation method of the average grayscale of the healthy area is as follows:

[0082] Among them, h2 represents the average grayscale of the healthy area, α2 represents the proportion of the healthy area in the image, B(m) represents the proportion of grayscale m in the image, m represents the grayscale, and b represents the image segmentation into b categories.

[0083] The feature extraction module is used to receive the data and images analyzed and calculated in the cucumber leaf image data analysis and calculation module, and extract features from the data and images;

[0084] In the feature extraction module, the three component features in the HIS model are extracted, and the method for converting RGB to HIS is specifically as follows:

[0085] The calculation method of hue value is as follows:

[0086] Among them, H represents the hue value, R represents the red value in the RGB model, G represents the green value in the RGB model, and B represents the blue value in the RGB model;

[0087] The specific calculation method of brightness value is:

[0088] Among them, I represents the brightness value, R represents the red value in the RGB model, G represents the green value in the RGB model, and B represents the blue value in the RGB model;

[0089] The specific calculation method of saturation is:

[0090] Among them, S represents saturation, R represents the red value in the RGB model, G represents the green value in the RGB model, and B represents the blue value in the RGB model;

[0091] The calculation method of texture feature value is as follows:

[0092] Where W represents the texture eigenvalue, N(l, θ) represents the energy value in the texture eigenvalue, C(l, θ) represents the entropy value in the texture eigenvalue, and G(l, θ) represents the moment of inertia value in the texture eigenvalue;

[0093] The specific method for calculating the energy value in the texture feature is:

[0094] N(l,θ)=∑ i，j K(i, j|l, θ) 2 , where N(l, θ) represents the energy value in the texture feature value, K(i, j|l, θ) represents the element value, (i, j) represents the probability of the grayscale layer appearing, l represents the distance, and θ represents the direction angle;

[0095] The specific method for calculating the entropy value in texture features is:

[0096] C(l,θ)=∑ i，j K(i, j|l, θ)-logK(i, j|l, θ), where C(l, θ) represents the entropy value in the texture feature value, K(i, j|l, θ) represents the element value, (i, j) represents the probability of the grayscale layer appearing, l represents the distance, and θ represents the direction angle;

[0097] The calculation method of the moment of inertia value in the texture feature is as follows:

[0098] G(l,θ)=∑ i，j (ij) 2K(i, j|l, θ), where G(l, θ) represents the moment of inertia value in the texture feature value, K(i, j|l, θ) represents the element value, (i, j) represents the probability of occurrence of the grayscale layer, l represents the distance, and θ represents the direction angle.

[0099] The disease recognition neural network model establishment and training module is used to establish a disease recognition model, and train the model to classify and identify different diseases using feature-extracted data;

[0100] In the disease recognition neural network model establishment and training module, a disease recognition neural network model is established, and the disease value calculation method is specifically as follows:

[0101] Among them, Z represents the disease value, H represents the hue value, I represents the brightness value, S represents the saturation, W represents the texture feature value, N(l, θ) represents the energy value in the texture feature value, C(l, θ) represents the entropy value in the texture feature value, and G(l, θ) represents the moment of inertia value in the texture feature value;

[0102] The specific method of constructing a neural network model for disease identification is as follows:

[0103] Select the BP neural network structure, and use the disease value, color feature value and texture feature value as the input value of the neural network, denoted as v1, v2, v3;

[0104] Input the input pattern v to the input layer neurons, and forward calculate the values ​​of the hidden layer neurons. The calculation method of the values ​​of the hidden layer neurons is as follows:

[0105] Among them, E1 represents the value of the hidden layer neuron, U represents the number of neurons in the input layer, d represents the dth neuron in the input and output layer, and DI od It is represented by the connection weights from the input layer to the hidden layer neurons determined after training, FI o It is represented by the threshold value of the input layer to the hidden layer neurons determined after training;

[0106] The output value of the hidden layer neuron is forward calculated. The calculation method of the value of the output layer neuron is as follows:

[0107] Among them, P a Represented as the value of the output layer neuron, represents the number of neurons in the output layer, o represents the oth neuron in the hidden layer, DT od It is represented by the connection weights from the hidden layer neurons to the output layer neurons determined after training, FI d It is represented as the threshold from hidden layer neurons to output layer neurons determined after training.

[0108] The output layer neuron value P calculated by the BP network a , a=1, 2, ..., X2, compare their sizes, and take the final output as neuron P max , where P max =max(t a ), a=1, 2, ..., X2, and the corresponding image category is the final classification result, and the final result is transmitted to the control end display module.

[0109] The system integration module is used to integrate the trained cucumber leaf disease recognition model into the computer image processing system;

[0110] The system integration module includes the following units:

[0111] Model export and deployment unit: First, you need to export the trained recognition model and save the model's weight parameters and structure information as a file; common formats include TensorFlow's .pb file and PyTorch's .pth file; deploy the model to the server according to the selected deployment method of the target system;

[0112] Interface design and data format processing unit: Design the interface for model calls to ensure that the system can interact with the model; unify the format requirements of input data, including the size and format of the image, and the format of the output results. The format of the output results refers to the classification results and disease location information; it is also necessary to handle data preprocessing and postprocessing to ensure that the input data meets the requirements of the model, and explain and display the output results;

[0113] Integration test and optimization unit: After integrating the model into the computer image processing system, conduct system-level integration testing to verify the performance of the model in actual application scenarios; optimize and adjust according to the test results to ensure that the overall performance of the system meets expectations;

[0114] Continuous update and maintenance unit: With the accumulation of disease data and the training of models, the system regularly updates the models to improve recognition accuracy. At the same time, it promptly handles errors and problems that arise in actual applications of the models to ensure the stability and reliability of the system.

[0115] The control end display module is used to identify the disease type in the disease model and send the disease type to the control end;

[0116] In the control end display module, a receiving port is set on the control end for receiving disease type information transmitted from the disease identification model, and the control end is a computer terminal; the received disease type information is displayed on the control end, and the type information is displayed in the form of graphics and text. The name, picture and corresponding diagnosis method of the disease type are displayed on the control end.

[0117] See also Figure 2 As shown, in this embodiment, it needs to be specifically explained that the present invention provides a computer data processing method based on big data analysis, comprising the following steps:

[0118] Step Q1: Use a shooting tool to shoot an image of a cucumber leaf under natural light conditions;

[0119] Step Q2: performing image processing on the collected cucumber leaf image;

[0120] Step Q3: Analyze and calculate the processed image data;

[0121] Step Q4: extracting features from the calculated image data;

[0122] Step Q5: Establishing a disease recognition neural network model and training the model;

[0123] Step Q6: Integrate the trained cucumber leaf disease recognition model into a computer image processing system;

[0124] Step Q7: Identify the disease type in the disease model and send the disease type to the control end.

[0125] The above steps are the same as those in the system module and will not be repeated here.

[0126] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A data recognition system based on a disease recognition model, characterized in that: include: Image data acquisition module: used to use a shooting tool to shoot cucumber leaf images under natural light conditions, and transmit the shot cucumber leaf images to the image data processing module; Cucumber leaf image data processing module: used to receive the cucumber leaf image collected by the image data collection module for image processing, and transmit the processed image to the cucumber leaf image data analysis and calculation module; Cucumber leaf image data analysis and calculation module: used to receive the processed image in the cucumber leaf image data processing module, and analyze and calculate the processed image data; Feature extraction module: used to receive the data and images analyzed and calculated in the cucumber leaf image data analysis and calculation module, and extract features of the data and images; Disease recognition neural network model establishment and training module: used to establish a disease recognition model and train the model to classify and identify different diseases using feature-extracted data; The specific method of constructing a neural network model for disease identification is as follows: Select the BP neural network structure, and use the disease value, color feature value and texture feature value as the input value of the neural network, denoted as v1, v2, v3; Input the input pattern v to the input layer neurons, and forward calculate the values ​​of the hidden layer neurons. The calculation method of the values ​​of the hidden layer neurons is as follows: Among them, E1 represents the value of the hidden layer neuron, U represents the number of neurons in the input layer, d represents the dth neuron in the input and output layer, and DI od It is represented by the connection weights from the input layer to the hidden layer neurons determined after training, FI o It is represented by the threshold value of the input layer to the hidden layer neurons determined after training; The output value of the hidden layer neuron is forward calculated. The calculation method of the value of the output layer neuron is as follows: Among them, P a Represented as the value of the output layer neuron, It represents the number of neurons in the output layer, 0 represents the 0th neuron in the hidden layer, DTo d It is represented by the connection weights from the hidden layer neurons to the output layer neurons determined after training, FI d It is represented by the threshold from hidden layer neurons to output layer neurons determined after training; System integration module: used to integrate the trained cucumber leaf disease recognition model into the computer image processing system; Control end display module: used to identify the disease type in the disease model and send the disease type to the control end.

2. The data identification system based on the disease identification model according to claim 1, characterized in that: In the image data acquisition module, the shooting tool is a digital camera. Under natural lighting conditions in the field, the automatic exposure mode is used to capture images of leaf diseases of cucumbers during the growth period with a resolution of 3840x2160 pixels, and the images are transmitted to the computer in BMP format.

3. The data identification system based on the disease identification model according to claim 1, characterized in that: In the cucumber leaf image data processing module, the collected image is subjected to image denoising by using Gaussian filtering.

4. The data identification system based on the disease identification model according to claim 1, characterized in that: The cucumber leaf image data analysis and calculation module performs grayscale processing on the processed image, converts it into a grayscale image, and then performs segmentation processing on the grayscale image.

5. The data identification system based on the disease identification model according to claim 1, characterized in that: In the feature extraction module, the three component features in the HIS model are extracted, and the method for converting RGB to HIS is specifically as follows: The calculation method of hue value is as follows: Among them, H represents the hue value, R represents the red value in the RGB model, G represents the green value in the RGB model, and B represents the blue value in the RGB model; The specific calculation method of brightness value is: Among them, I represents the brightness value, R represents the red value in the RGB model, G represents the green value in the RGB model, and B represents the blue value in the RGB model; The specific calculation method of saturation is: Among them, S represents saturation, R represents the red value in the RGB model, G represents the green value in the RGB model, and B represents the blue value in the RGB model; The calculation method of texture feature value is as follows: Among them, W represents the texture eigenvalue, N(l, θ) represents the energy value in the texture eigenvalue, C(l, θ) represents the entropy value in the texture eigenvalue, and G(l, θ) represents the moment of inertia value in the texture eigenvalue.

6. The data identification system based on the disease identification model according to claim 5, characterized in that: The specific method for calculating the energy value in the texture feature is: N(l, θ)=∑i,j K(i,j|l,θ) 2 , where N(l, θ) represents the energy value in the texture feature value, K(i, j|l, θ) represents the element value, (i, j) represents the probability of the grayscale layer appearing, l represents the distance, and θ represents the direction angle; The specific method for calculating the entropy value in texture features is: C(l, θ) = ∑i, j K(i, j|l, θ) - log K(i, j|l, θ), where C(l, θ) represents the entropy value in the texture feature value, K(i, j|l, θ) represents the element value, (i, j) represents the probability of the grayscale layer appearing, 1 represents the distance, and θ represents the direction angle; The calculation method of the moment of inertia value in the texture feature is as follows: G(l,θ)=∑i,j(ij) 2 K(i, j|l, θ), where G(l, θ) represents the moment of inertia value in the texture feature value, K(i, j|l, θ) represents the element value, (i, j) represents the probability of occurrence of the grayscale layer, l represents the distance, and θ represents the direction angle.

7. The data identification system based on the disease identification model according to claim 1, characterized in that: In the disease recognition neural network model establishment and training module, a disease recognition neural network model is established, and the disease value calculation method is specifically as follows: Among them, Z represents the disease value, H represents the hue value, I represents the brightness value, S represents the saturation, W represents the texture eigenvalue, N(l, θ) represents the energy value in the texture eigenvalue, C(l, θ) represents the entropy value in the texture eigenvalue, and G(l, θ) represents the moment of inertia value in the texture eigenvalue.

8. The data identification system based on the disease identification model according to claim 1, characterized in that: The system integration module includes the following units: Model export and deployment unit: First, you need to export the trained recognition model and save the model's weight parameters and structure information as a file; common formats include TensorFlow's .pb file and PyTorch's .pth file; Deploy the model to the server according to the selected deployment method of the target system; Interface design and data format processing unit: Design the interface for model calls to ensure that the system can interact with the model; unify the format requirements of input data, including the size and format of the image, and the format of the output results. The format of the output results refers to the classification results and disease location information; It is also necessary to deal with data pre-processing and post-processing to ensure that the input data meets the requirements of the model and to interpret and display the output results; Integration test and optimization unit: After integrating the model into the computer image processing system, system-level integration testing is performed to verify the performance of the model in actual application scenarios; Optimize and adjust according to the test results to ensure that the overall system performance meets expectations; Continuous update and maintenance unit: With the accumulation of disease data and the training of models, the system regularly updates the models to improve recognition accuracy. At the same time, it promptly handles errors and problems that arise in actual applications of the models to ensure the stability and reliability of the system.

9. The data identification system based on the disease identification model according to claim 1, characterized in that: In the control end display module, a receiving port is set on the control end for receiving disease type information transmitted from the disease identification model, and the control end is a computer terminal; the received disease type information is displayed on the control end, and the type information is displayed in the form of graphics and text. The name, picture and corresponding diagnosis method of the disease type are displayed on the control end.

Images