Crop disease and pest identification method and device based on deep learning

Abstract

The invention belongs to the technical field of crop disease and insect pest recognition, and provides a crop disease and insect pest recognition method and device based on deep learning, and the method comprises the steps: S1, collecting image data of crops in a target region, and carrying out the preprocessing of the image data, and obtaining target data; s2, performing matching to obtain environment data corresponding to the target data; s3, generating standard input data based on the image historical data and the corresponding environment data, and training an artificial intelligence model through the standard input data and the standard output data; s4, integrating the target data and the corresponding environment data into a pest and disease identification sequence; according to the method, the image data are considered, the environment data such as the environment temperature and the environment humidity are introduced, the environment data corresponding to the target data are obtained through matching, information of more dimensions is provided for accurate recognition of diseases and insect pests, and therefore the recognition accuracy and reliability are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of crop pest and disease identification, and specifically relates to a method and device for crop pest and disease identification based on deep learning. Background Technique

[0002] In the agricultural field, the accurate identification and timely prevention of crop pests and diseases are crucial for ensuring crop yield and quality. However, traditional pest and disease identification methods mainly rely on manual experience, suffering from problems such as low identification efficiency, insufficient accuracy, and difficulty in large-scale application. With the rapid development of artificial intelligence and deep learning technologies, remarkable achievements have been made in the field of image recognition and classification, providing new solutions for the automatic identification of crop pests and diseases.

[0003] Currently, although some crop pest and disease identification methods based on deep learning have been proposed, these methods often rely solely on the image data itself, ignoring the impact of environmental factors (including temperature, humidity, etc.) on the occurrence and development of pests and diseases. In addition, these methods still need to be improved in terms of identification accuracy and generalization ability. Especially in the complex and changeable farmland environment, their identification effects are often greatly limited.

[0004] Therefore, those skilled in the art have proposed a method and device for crop pest and disease identification based on deep learning to solve the problems raised in the background technique. Summary of the Invention

[0005] To solve the above technical problems, the present invention provides a method and device for crop pest and disease identification based on deep learning to solve the problems in the prior art that pest and disease identification methods mainly rely on manual experience, with low identification efficiency, insufficient accuracy, and difficulty in large-scale application.

[0006] A method for crop pest and disease identification based on deep learning includes:

[0007] S1. First, collect image data of crops in the target area and preprocess the image data to obtain target data;

[0008] S2. Match and obtain the environmental data corresponding to the target data, where the environmental data includes environmental temperature, environmental humidity, and collection time; Matching and obtaining the environmental data corresponding to the target data includes: extracting the collection time corresponding to the target data; based on the collection time, matching the weather data at the corresponding time from the database, and integrating the weather data and the collection time into the environmental data of the target data;

[0009] S3. Generate standard input data based on the image historical data and the corresponding environmental data, and train an artificial intelligence model through the standard input data and standard output data to obtain a pest and disease identification model;

[0010] S4. Integrate the target data and the corresponding environmental data into a pest and disease identification sequence, combine it with the pest and disease identification model to obtain the corresponding pest and disease identification result, and mark it as the original result;

[0011] S5. Take the environmental temperature and environmental humidity as independent variables, and take the difference between the pest and disease coverage rate corresponding to the image historical data and the actual pest and disease coverage rate as the dependent variable to establish an identification correction model; among them, the actual pest and disease coverage rate refers to the true coverage rate of crop pests and diseases in the actual farmland environment. Specifically, it represents the proportion or degree of actual damage to crops by pests and diseases;

[0012] S6. Correct the original result based on the identification correction model to obtain the final identification result; correcting the original result based on the identification correction model includes: extracting the environmental temperature and environmental humidity corresponding to the original result and importing them into the correction identification model to obtain a correction coefficient; superimposing the correction coefficient and the original result to obtain the final identification result.

[0013] Preferably, in step S1, in the image preprocessing step, in order to enhance the local contrast of the image, a contrast-limited adaptive histogram equalization (CLAHE) algorithm is introduced. CLAHE effectively avoids the problem of over-enhancement by restricting the amplitude of local contrast enhancement.

[0014] Preferably, in step S2, a weighted fusion algorithm is adopted for the environmental data to fuse multi-source data.

[0015] Preferably, in step S3, in generating the standard input data based on the image historical data and the corresponding environmental data, in order to obtain richer texture information in the image, a local binary pattern (LBP) algorithm is introduced. LBP is an effective texture description operator that can capture the local structure information of the image.

[0016] Preferably, in step S3, the artificial intelligence model adopts a convolutional neural network (CNN). The convolutional neural network (CNN) includes:

[0017] Convolutional layer, which extracts local features of the input data through convolutional operations;

[0018] Pooling layer, which is used to reduce the spatial dimension of the feature map, reduce the amount of calculation, and retain important features at the same time.

[0019] Preferably, in step S3, in training the artificial intelligence model, a backpropagation algorithm is introduced, which calculates the gradient of the loss function with respect to the model parameters and updates the parameters to minimize the loss function.

[0020] Preferably, in step S4, after obtaining the corresponding pest and disease identification result by combining the pest and disease identification model, in order to optimize the spatial consistency of the result, a conditional random field (CRF) model can be introduced for post-processing. CRF is a probabilistic graphical model used to model the dependencies between label sequences.

[0021] A device for identifying pests and diseases of crops based on deep learning, using the above-mentioned method for identifying pests and diseases of crops based on deep learning, includes:

[0022] An image acquisition module for acquiring image data of crops in the target area;

[0023] A data preprocessing module for preprocessing the acquired image data to obtain target data;

[0024] An environmental data matching module for matching and obtaining environmental data corresponding to the target data;

[0025] A model training module for generating standard input data based on historical image data and corresponding environmental data, training an artificial intelligence model through the standard input data and standard output data, and obtaining a pest and disease identification model;

[0026] An identification result generation module for integrating the target data and corresponding environmental data into a pest and disease identification sequence, obtaining the corresponding pest and disease identification result by combining the pest and disease identification model, and marking it as the original result;

[0027] An identification result correction module for establishing an identification correction model with environmental temperature and environmental humidity as independent variables and the difference between the pest and disease coverage rate corresponding to the historical image data and the actual pest and disease coverage rate as the dependent variable, and correcting the original result based on the identification correction model to obtain the final identification result.

[0028] A processor configured to execute a method for identifying pests and diseases of crops based on deep learning as described above.

[0029] A computer-readable storage medium having a computer program stored thereon, where the computer program, when executed by a processor, implements the above-mentioned method for identifying pests and diseases of crops based on deep learning.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] 1. In the identification process, the present invention not only considers image data, but also introduces environmental data such as environmental temperature and environmental humidity. By matching and obtaining environmental data corresponding to the target data, more dimensional information is provided for the accurate identification of pests and diseases, thereby improving the accuracy and reliability of the identification.

[0032] 2. The present invention takes the ambient temperature and ambient humidity as independent variables, and takes the difference between the pest and disease coverage rate corresponding to the image historical data and the actual pest and disease coverage rate as the dependent variable, and establishes an identification correction model; this model can correct the preliminary identification result and further improve the identification accuracy.

[0033] 3. The present invention adopts advanced deep learning algorithms such as convolutional neural network (CNN) in the model training process, which can automatically extract the feature information in the image, thereby improving the identification ability and generalization ability of the model.

[0034] 4. After the identification result is generated, the present invention can also introduce a conditional random field (CRF) model for post-processing to optimize the spatial consistency of the result, so that the identification result is more in line with the actual situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a flowchart of the method for identifying crop pests and diseases based on deep learning of the present invention;

[0036] Figure 2 is a framework diagram of the device for identifying crop pests and diseases based on deep learning of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] The following further describes the embodiments of the present invention in detail with reference to the drawings and embodiments. The following embodiments are used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0038] Embodiment: The present invention provides a method for identifying crop pests and diseases based on deep learning, as Figure 1 shown, including:

[0039] S1. First, collect the image data of the crops in the target area, and preprocess the image data to obtain the target data;

[0040] S2. Match and obtain the environmental data corresponding to the target data, where the environmental data includes the ambient temperature, ambient humidity, and collection time; matching and obtaining the environmental data corresponding to the target data includes: extracting the collection time corresponding to the target data; based on the collection time, matching the weather data at the corresponding time from the database, and integrating the weather data and the collection time into the environmental data of the target data;

[0041] S3. Generate standard input data based on the image historical data and the corresponding environmental data, and train the artificial intelligence model with the standard input data and the standard output data to obtain a pest and disease identification model;

[0042] S4. Integrate the target data and the corresponding environmental data into a pest and disease identification sequence, combine it with the pest and disease identification model to obtain the corresponding pest and disease identification result, and mark it as the original result;

[0043] S5. Taking the ambient temperature and ambient humidity as independent variables, and taking the difference between the pest and disease coverage rate corresponding to the image historical data and the actual pest and disease coverage rate as the dependent variable, an identification and correction model is established. When establishing the identification and correction model, a linear regression algorithm is introduced to establish the correction model. The formula of the linear regression algorithm includes:

[0044] Y = β0 + β1X1 + β2X2 +... + β n X n + ε;

[0045] Among them, Y represents the correction coefficient, X1, X2,..., X n represents the independent variables (including ambient temperature, ambient humidity), β0, β1,..., β n represents the regression coefficients, and ε represents the error term;

[0046] S6. Based on the identification and correction model, the original result is corrected to obtain the final identification result. Correcting the original result based on the identification and correction model includes: extracting the ambient temperature and ambient humidity corresponding to the original result and importing them into the correction and identification model to obtain the correction coefficient; superimposing the correction coefficient and the original result to obtain the final identification result.

[0047] As can be seen from the above, this method not only considers the image data, but also introduces environmental data such as ambient temperature and ambient humidity. By matching to obtain the environmental data corresponding to the target data, it provides more dimensional information for the accurate identification of pests and diseases, thereby improving the accuracy and reliability of the identification; in addition, by establishing an identification and correction model to correct the preliminary identification result, the identification accuracy is further improved; at the same time, using advanced deep learning algorithms such as convolutional neural networks for model training can automatically extract the feature information in the image and continuously optimize the model parameters through the backpropagation algorithm, improving the identification ability and generalization ability of the model; finally, by introducing a conditional random field model for post-processing, the spatial consistency of the result is optimized, making the identification result more in line with the actual situation.

[0048] Furthermore, in step S1, in the image preprocessing step, in order to enhance the local contrast of the image, a contrast-limited adaptive histogram equalization (CLAHE) algorithm is introduced. CLAHE effectively avoids the problem of over-enhancement by restricting the amplitude of local contrast enhancement. The formula of the adaptive histogram equalization (CLAHE) algorithm includes:

[0049]

[0050] Among them, x, y are the coordinates of the image pixels, CDF(x, y) is the cumulative distribution function (CDF) value of the local area where the pixel is located, CDF max and CDFmin They are respectively the minimum and maximum values of the CDF of the local region. α and β are parameters for adjusting contrast and brightness, and clip(·) is a clipping function used to limit the result within a valid range.

[0051] As can be seen from the above, the contrast-limited adaptive histogram equalization (CLAHE) algorithm is introduced in the image preprocessing step. By limiting the amplitude of local contrast enhancement, this CLAHE algorithm effectively avoids the problem of over-enhancement of the image, thereby enhancing the local contrast of the image and making the detailed information in the image clearer and more distinguishable. This processing not only improves the quality of the image but also helps to more accurately extract image features in subsequent steps, thus improving the accuracy and reliability of the entire crop pest and disease identification method.

[0052] Furthermore, in step S2, a weighted fusion algorithm is adopted for environmental data to fuse multi-source data. The formula of the weighted fusion algorithm is as follows:

[0053]

[0054] Among them, F is the fused feature, F i is the feature of the i-th data source, α i is the corresponding weight coefficient, and N is the number of data sources.

[0055] As can be seen from the above, a weighted fusion algorithm is adopted for environmental data to fuse multi-source data. By assigning corresponding weight coefficients to different data sources, this algorithm realizes the comprehensive consideration of environmental data and effectively integrates the influence of various environmental factors (including environmental temperature, environmental humidity, etc.) on the occurrence of crop pests and diseases. This fusion method not only improves the comprehensiveness and accuracy of environmental data but also provides a richer and more reliable information basis for subsequent pest and disease identification, thereby further enhancing the accuracy and reliability of pest and disease identification.

[0056] Furthermore, in step S3, in generating standard input data based on image historical data and corresponding environmental data, in order to obtain richer texture information in the image, the local binary pattern (LBP) algorithm is introduced. LBP is an effective texture description operator that can capture the local structural information of the image. The formula of the local binary pattern (LBP) algorithm includes:

[0057]

[0058] Among them, (x c , y c ) is the coordinate of the central pixel, g c is the gray value of the central pixel, g pis the gray value of its neighboring pixels, P is the number of neighboring pixels (usually 8), and s(·) is the sign function, defined as:

[0059]

[0060] As can be seen from the above, introducing the Local Binary Pattern (LBP) algorithm to obtain richer texture information in the image can accurately capture the local structural information of the image, providing more detailed image features for subsequent pest and disease identification; by introducing the LBP algorithm, the standard input data not only contains the basic information of the image but also incorporates rich texture details, which helps the pest and disease identification model to more accurately identify the characteristics of different pests and diseases, improving the accuracy and generalization ability of the identification, and making the identification of crop pests and diseases more accurate and reliable.

[0061] Furthermore, in step S3, the artificial intelligence model adopts a Convolutional Neural Network (CNN), and the Convolutional Neural Network (CNN) includes:

[0062] The convolutional layer extracts the local features of the input data through convolutional operations, and its formula includes:

[0063]

[0064] where I is the input image, K is the convolutional kernel, (i,j) are the pixel coordinates on the output feature map, and (m,n) is the size of the convolutional kernel;

[0065] The pooling layer is used to reduce the spatial dimension of the feature map, reduce the computational amount, and at the same time retain important features, and its formula includes:

[0066]

[0067] where P(i,j) is the pixel value after pooling, I(m,n) is the pixel value on the input feature map, and R ij is the position of the pooling window on the feature map.

[0068] As can be seen from the above, using the Convolutional Neural Network (CNN) as the artificial intelligence model, its convolutional layer can effectively extract the local features of the input data through convolutional operations, and these features are crucial for the identification of pests and diseases; the pooling layer further reduces the spatial dimension of the feature map, reduces the computational amount, and at the same time retains important features, improving the computational efficiency of the model; this combined structure enables the CNN model to automatically learn and identify the characteristic patterns of crop pests and diseases without the need for artificial design of complex feature extractors, greatly improving the accuracy and efficiency of the identification, and providing strong technical support for the automatic identification of crop pests and diseases.

[0069] Further, in step S3, when training the artificial intelligence model, the backpropagation algorithm is introduced. It calculates the gradient of the loss function with respect to the model parameters and updates the parameters to minimize the loss function. The formula of the backpropagation algorithm includes:

[0070]

[0071] where W (l) is the weight matrix of the l-th layer, η is the learning rate, L is the loss function, δ (l+1) is the error term of the (l + 1)-th layer, and a (l) is the activation value of the l-th layer.

[0072] As can be seen from the above, when training the artificial intelligence model, the backpropagation algorithm is introduced. This algorithm accurately calculates the gradient of the loss function with respect to the model parameters and updates the model parameters according to the gradient information to minimize the loss function, thereby continuously optimizing the performance of the model, realizing the automatic adjustment of the model parameters, and enabling the convolutional neural network (CNN) model to gradually learn more accurate pest and disease feature representations. This automated optimization method not only improves the training efficiency of the model but also significantly enhances the recognition accuracy and generalization ability of the model, ensuring the stability and reliability of the crop pest and disease recognition method in practical applications.

[0073] Further, in step S4, after obtaining the corresponding pest and disease recognition result by combining with the pest and disease recognition model, in order to optimize the spatial consistency of the result, a conditional random field (CRF) model can be introduced for post-processing. CRF is a probabilistic graphical model for modeling the dependencies between label sequences. The formula of the conditional random field (CRF) model includes:

[0074]

[0075] where y is the label sequence, Ψ u (y i ) is the unary potential function (usually related to the preliminary recognition result), and Ψ p (y i , y j ) is the pairwise potential function (used to model the dependencies between labels); by minimizing the energy function, the optimal label sequence can be obtained.

[0076] As can be seen from the above, introducing the Conditional Random Field (CRF) model for post-processing can model the dependencies between label sequences. By considering the mutual influence between adjacent labels, the spatial consistency of the pest and disease identification results is effectively optimized. This post-processing step not only improves the accuracy of the identification results but also makes the identification results more in line with the actual situation of the distribution of crop pests and diseases in the field. By minimizing the energy function, the CRF model can find the optimal label sequence, further enhancing the reliability and practicality of pest and disease identification, and providing strong support for the precise prevention and control of agricultural pests and diseases.

[0077] Furthermore, a deep learning-based crop pest and disease identification method of the embodiment was compared with the current deep learning-based crop pest and disease identification method (comparative example), and the following table was obtained:

[0078]

[0079]

[0080] As can be seen from the above table, a deep learning-based crop pest and disease identification method of the embodiment shows advantages in data fusion, identification accuracy, generalization ability, environmental adaptability, post-processing optimization, and practical applications. It comprehensively considers image data and environmental factors, improves the identification accuracy by introducing an identification correction model and conditional random field post-processing, has strong generalization ability and environmental adaptability, is more suitable for complex and changeable farmland environments, and has broad application prospects.

[0081] A deep learning-based crop pest and disease identification device, as Figure 2 shown, uses the above deep learning-based crop pest and disease identification method, including:

[0082] An image acquisition module for acquiring image data of crops in the target area;

[0083] A data preprocessing module for preprocessing the acquired image data to obtain target data;

[0084] An environmental data matching module for matching and obtaining environmental data corresponding to the target data;

[0085] A model training module for generating standard input data based on historical image data and corresponding environmental data, training an artificial intelligence model through the standard input data and standard output data to obtain a pest and disease identification model;

[0086] An identification result generation module for integrating the target data and corresponding environmental data into a pest and disease identification sequence, obtaining a corresponding pest and disease identification result in combination with the pest and disease identification model, and marking it as the original result;

[0087] An identification result correction module, which uses the ambient temperature and ambient humidity as independent variables, and the difference between the pest and disease coverage rate corresponding to the image historical data and the actual pest and disease coverage rate as the dependent variable, establishes an identification correction model, and corrects the original result based on the identification correction model to obtain the final identification result.

[0088] Working principle: First, the image acquisition module acquires the image data of the crops in the target area, and the data preprocessing module performs preprocessing to obtain the target data; then, the environmental data matching module matches and obtains the environmental data corresponding to the target data, including the ambient temperature, ambient humidity, and acquisition time; then, based on the image historical data and its corresponding environmental data, standard input data is generated, and a convolutional neural network (CNN) model is trained using these data to obtain a pest and disease identification model; in the identification stage, the target data and environmental data are integrated into a pest and disease identification sequence and input into the pest and disease identification model to obtain a preliminary identification result; in order to further improve the identification accuracy, an identification correction model is established with the ambient temperature and ambient humidity as independent variables and the difference between the pest and disease coverage rate corresponding to the image historical data and the actual pest and disease coverage rate as the dependent variable, and the preliminary identification result is corrected to obtain the final identification result; in addition, a conditional random field (CRF) model can be introduced to post-process the identification result to optimize the spatial consistency of the result; the entire technical solution realizes the accurate identification of crop pests and diseases by comprehensively considering the image data and environmental data and combining advanced deep learning algorithms.

[0089] An embodiment of the present application provides an electronic device, which is applicable to the above-mentioned method for identifying crop pests and diseases based on deep learning, including:

[0090] A memory for storing computer programs and data;

[0091] A processor for running the system program.

[0092] An embodiment of the present application provides a computer storage medium, which is applicable to the above-mentioned method for identifying crop pests and diseases based on deep learning, and performs hierarchical confidentiality management on the above-mentioned system and data according to the confidentiality management requirements.

[0093] Those skilled in the art should understand that the embodiments of the present application can be provided as a system or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0094] This application is described with reference to the flowcharts and / or block diagrams of devices (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices produce a means for implementing the functions specified in the Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0095] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction means that implements the functions specified in the Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0096] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in the Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0097] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0098] The memory includes non-permanent memory in the form of computer-readable media, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. The memory is an example of computer-readable media.

[0099] Computer readable media include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include transitory media such as modulated data signals and carrier waves.

[0100] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, commodity or device including the elements.

[0101] The embodiments of the present invention are provided for the purpose of illustration and description. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations of the present invention. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for identifying crop diseases and insect pests based on deep learning, characterized in that: include: S1. First, image data of crops in the target area are collected, and the image data are preprocessed to obtain target data; S2, matching and obtaining environmental data corresponding to the target data, wherein the environmental data includes environmental temperature, environmental humidity and collection time; the matching and obtaining environmental data corresponding to the target data includes: extracting the collection time corresponding to the target data; matching the weather data of the corresponding time from the database based on the collection time, and integrating the weather data and the collection time into the environmental data of the target data; S3, generating standard input data based on image history data and corresponding environmental data, training an artificial intelligence model through the standard input data and standard output data, and obtaining a pest and disease recognition model; S4, integrating the target data and the corresponding environmental data into a pest and disease identification sequence, combining the pest and disease identification model to obtain the corresponding pest and disease identification result, and marking it as the original result; S5. Taking the ambient temperature and ambient humidity as independent variables, and taking the difference between the pest and disease coverage rate corresponding to the image history data and the actual pest and disease coverage rate as the dependent variable, establish an identification and correction model; S6. Correcting the original result based on the recognition correction model to obtain a final recognition result; correcting the original result based on the recognition correction model includes: extracting the ambient temperature and ambient humidity corresponding to the original result, and importing them into the corrected recognition model to obtain a correction coefficient; superimposing the correction coefficient and the original result to obtain a final recognition result.

2. A method for identifying crop pests and diseases based on deep learning as claimed in claim 1, characterized in that: In step S1, in the image preprocessing step, in order to enhance the local contrast of the image, a contrast-limited adaptive histogram equalization algorithm is introduced.

3. A method for identifying crop diseases and insect pests based on deep learning as claimed in claim 1, characterized in that: In step S2, a weighted fusion algorithm is used for the environmental data.

4. The method for identifying crop pests and diseases based on deep learning as claimed in claim 1, characterized in that: In step S3, in generating standard input data based on image history data and corresponding environmental data, a local binary pattern algorithm is introduced to obtain richer texture information in the image.

5. The method for identifying crop pests and diseases based on deep learning as claimed in claim 1, characterized in that: In step S3, the artificial intelligence model adopts a convolutional neural network, and the convolutional neural network includes: Convolutional layer, extracts local features of input data through convolution operation; The pooling layer is used to reduce the spatial dimension of the feature map, reduce the amount of computation, and retain important features.

6. The method for identifying crop pests and diseases based on deep learning as claimed in claim 1, characterized in that: In step S3, a back propagation algorithm is introduced in the training of the artificial intelligence model, which calculates the gradient of the loss function with respect to the model parameters and updates the parameters to minimize the loss function.

7. The method for identifying crop pests and diseases based on deep learning as claimed in claim 1, characterized in that: In step S4, after obtaining the corresponding pest and disease identification result in combination with the pest and disease identification model, a conditional random field model may be introduced for post-processing in order to optimize the spatial consistency of the result.

8. A crop pest identification device based on deep learning, characterized in that: A method for identifying crop pests and diseases based on deep learning according to any one of claims 1 to 7, comprising: An image acquisition module, used to collect image data of crops in a target area; A data preprocessing module is used to preprocess the collected image data to obtain target data; An environmental data matching module is used to match and obtain environmental data corresponding to target data; A model training module is used to generate standard input data based on image history data and corresponding environmental data, train an artificial intelligence model through standard input data and standard output data, and obtain a pest and disease recognition model; The recognition result generation module is used to integrate the target data and the corresponding environmental data into a pest and disease recognition sequence, combine the pest and disease recognition model to obtain the corresponding pest and disease recognition result, and mark it as the original result; The recognition result correction module is used to establish a recognition correction model with ambient temperature and ambient humidity as independent variables and the difference between the pest and disease coverage rate corresponding to the image historical data and the actual pest and disease coverage rate as the dependent variable, and to correct the original result based on the recognition correction model to obtain the final recognition result.

9. A processor, characterized in that: The method is configured to execute a crop disease and insect pest identification method based on deep learning according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the method for identifying crop diseases and pests based on deep learning as described in any one of claims 1 to 7 is implemented.

Images