Evaluation method and evaluation system for pest control effect of botanical pesticide

Abstract

The invention discloses a method and system for evaluating the pest control effect of a botanical pesticide, and relates to the technical field of pest control, and the method comprises the steps: obtaining the image data of plant pests in a target control region, extracting the symptoms of pests, and recognizing the types of pests; screening prevention and control data meeting the standard by utilizing the prevention and control effect, and constructing a prevention and control state model of the disease and pest category under the standard prevention and control effect through a multivariate state estimation method to obtain reference probability distribution under the standard prevention and control effect; importing the plant disease and insect pest image data after the application of the botanical pesticide into a prevention and control state model to obtain real-time residual data, and obtaining probability distribution of a current prevention and control state after the application of the botanical pesticide based on the real-time residual data; and comparing the probability distribution of the current prevention and control state with the reference probability distribution to evaluate the prevention and control effect. According to the method, the control effect of the botanical pesticide is rapidly and accurately evaluated through multivariate state estimation, and a data basis is provided for screening of the botanical pesticide in the pest control process.

Description

Technical Field

[0001] The present invention relates to the technical field of pest control, and more specifically, to a method and system for evaluating the pest control effect of botanical pesticides. Background Art

[0002] Botanical pesticides are pesticides made from plant parts or extracts containing active ingredients that contain insecticides and fungicides. As biopesticides, botanical pesticides offer unparalleled advantages over traditional chemical pesticides, including low toxicity, easy degradation, and environmental friendliness. Recent research has seen significant progress. Based on their target, botanical pesticides can be categorized as botanical fungicides (such as botanical antibacterials, antifungals, and antivirals), botanical insecticides, botanical nematicides, and botanical herbicides. Active ingredients primarily include terpenes, alkaloids, flavonoids, essential oils, steroids, photoactivated toxins, toxic proteins, and tannins. Botanical pesticides are widely used in all aspects of agricultural production.

[0003] In targeted pest control, it is necessary to measure the antibacterial activity of botanical pesticides. However, due to unstable environmental conditions, data processing is difficult, the processing time is long, and the damage symptoms are not intuitive, which is not conducive to quickly and accurately evaluating the pest control effects of botanical pesticides. Therefore, there is currently a lack of accurate and rapid evaluation methods for the control effects of botanical pesticides. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention proposes a method and system for evaluating the pest control effect of botanical pesticides, which can achieve rapid and accurate evaluation of the control effect of botanical pesticides through multivariate state estimation, and provide a data basis for the screening of botanical pesticides in the process of pest control.

[0005] The present invention provides a method for evaluating the pest control effect of a botanical pesticide, comprising the following steps:

[0006] Acquire plant disease and insect pest image data within the target control area, pre-process the collected plant disease and insect pest image data, extract damage symptoms and identify the disease and insect pest categories in the plant disease and insect pest image data;

[0007] Retrieving control data according to the pest and disease category, and screening the control data that meets the standards using the control effect, constructing a control state model of the pest and disease category under the standard control effect using a multivariate state estimation method, and using the control state model to obtain a baseline probability distribution under the standard control effect;

[0008] Importing plant pest and disease image data after application of botanical pesticides into the control status model to obtain real-time residual data, and obtaining a probability distribution of the current control status after application of the botanical pesticide based on the real-time residual data;

[0009] By comparing the probability distribution of the current control status with the baseline probability distribution, the control effect is evaluated to obtain the control effect evaluation results of the diseases and pests of plants in the target control area by applying botanical pesticides.

[0010] In this solution, we obtain plant disease and insect pest image data within the target control area, pre-process the collected plant disease and insect pest image data, extract damage symptoms and identify the disease and insect pest categories in the plant disease and insect pest image data, specifically:

[0011] Acquire plant pest and disease images in the target control area through visual equipment, adaptively preprocess the acquired plant pest and disease images according to the lighting environment, and import the preprocessed plant pest and disease images into the pest and disease recognition network based on the U-Net architecture;

[0012] Using the encoder of the pest and disease recognition network to extract features from plant pest and disease images, the encoder replaces the original structure with an AlexNet convolutional neural network with the fully connected layer removed, extracts feature maps through convolutional layers with convolution kernels of different sizes, and uses L2 regularization for weight decay after each convolution layer to minimize overfitting;

[0013] The extracted feature map is activated by the ELU activation function and input into the pooling layer, where the feature map is downsampled to reduce the dimension of the feature map for further feature extraction;

[0014] The feature map after the pooling operation is imported into the decoder, and the feature map is restored through upsampling and skip connection to obtain the detailed features in the feature map. The feature map after each layer is restored is subjected to binary classification segmentation to obtain the pest and disease damage area;

[0015] The pest and disease infested area is used as local information to generate recognition and prediction branches, and the various recognition and prediction branches are fused to achieve global information integration. The damage symptoms of plant pests and diseases are obtained through the global information, and the Sigmoid layer is used to obtain the probability distribution of the pest and disease category label, and the pest and disease category corresponding to the plant in the target prevention and control area is output.

[0016] In this solution, control data are retrieved according to the pest category, and control data that meets the standards are screened using control effects, specifically:

[0017] Establish search tags based on the acquired pest and disease categories and plant categories, use big data methods to obtain different types of control data based on the search tags, and preliminarily screen the control data based on the current growth stage of the plants in the target control area to obtain control data consistent with the growth stage;

[0018] The disease index of different control methods before and after control is extracted from the preliminarily screened control data, the control effect of each control data is obtained using the disease index, and a control effect threshold is obtained based on the growth condition of the plant. The control data with a control effect greater than the control effect threshold is screened, and plant images after control are extracted from the screened control data. A standard control effect image set of plants is constructed to characterize the control status of the pest and disease category under the standard control effect.

[0019] In this scheme, a multivariate state estimation method is used to construct a control state model of the current pest and disease category under standard control efficacy. The control state model is used to obtain the benchmark probability distribution under standard control efficacy, specifically:

[0020] Extracting training samples based on a standard plant pest control effect image set, modeling the training samples using a multivariate state estimation method, selecting plant image samples from the standard plant pest control effect image set using a k-nearest neighbor algorithm, and obtaining k training samples with the closest distance to the observation vector corresponding to the input training sample;

[0021] Using the obtained k training samples to construct a memory matrix of the prevention and control state model, using the memory matrix to calculate the estimated vector of the input training sample, and training the prevention and control state model by replacing the input training sample;

[0022] Obtain the residuals of the observation vector and the estimated vector corresponding to the training sample, calculate the mean and variance of the residuals, and obtain the baseline probability distribution of the pest and disease category under the standard control efficacy based on the residual mean and variance.

[0023] In this solution, the plant disease and insect pest image data after the application of botanical pesticides is imported into the control state model to obtain real-time residual data. Based on the real-time residual data, the probability distribution of the current control state after the application of botanical pesticides is obtained, specifically:

[0024] Obtaining an optimal sampling time after applying the botanical pesticide in the target control area according to the mechanism of action of the botanical pesticide, and obtaining corresponding plant disease and insect pest image data based on the optimal sampling time;

[0025] Importing the plant disease and insect pest image data after the application of the botanical pesticide into the control state model, obtaining an observation vector corresponding to the plant disease and insect pest image data after the application of the botanical pesticide, and constructing a memory matrix based on the observation vector;

[0026] An estimated vector is calculated using the memory matrix, and real-time residual data output by the control status model is obtained based on the comparison between the estimated vector and the observed vector. The mean and variance of the real-time residual data are calculated to obtain the probability distribution of the corresponding control status after the application of the botanical pesticide.

[0027] In this scheme, the control effect is evaluated by comparing the probability distribution of the current control status with the baseline probability distribution, and the control effect evaluation results of the botanical pesticide on the plants in the target control area are obtained. Specifically:

[0028] Obtain the deviation value between the probability distribution of the control state corresponding to the botanical pesticide and the baseline probability distribution, read the maximum component of the deviation value, and map the maximum component to the (0, 1) interval using the sigmoid function as the deviation degree;

[0029] The control effect of botanical pesticides is evaluated based on the deviation degree, the botanical pesticide control effect level and the corresponding numerical range of each level are preset, the numerical range into which the deviation degree falls is determined, the corresponding control effect level is read according to the numerical range into which it falls, and the final evaluation result is output.

[0030] In this solution, when the pest control effect level in the target control area after the application of botanical pesticides is less than the preset level threshold, the disease characteristics of the current pest damage are determined based on the collected plant pest image data;

[0031] Performing similarity calculation on the screened prevention and control data according to the disease characteristics, obtaining prevention and control data that meets the similarity requirements, determining the prevention and control measure categories corresponding to the prevention and control data, and screening green prevention and control measures according to the prevention and control measure categories;

[0032] A pest and disease epidemic duration prediction module is constructed based on a multi-layer perceptron. A corresponding number of prediction heads are set according to the quantity information of the green control measures. Model training is performed based on the control data corresponding to the selected green control measures.

[0033] The current disease characteristics of pests and diseases are used as model input to obtain the duration of pest and disease epidemics in the target prevention and control area under different green control measures. The green control measure with the shortest duration of pest and disease epidemics is selected for output, and a comprehensive prevention and control plan is generated in combination with botanical pesticides.

[0034] The second aspect of the present invention provides a plant-derived pesticide pest control effect evaluation system, comprising: a data acquisition module, a pest identification module, an emulation evaluation module and an evaluation result output module;

[0035] The data acquisition module is responsible for acquiring plant disease and insect pest image data within the target control area and preprocessing the acquired plant disease and insect pest image data;

[0036] The pest and disease identification module is responsible for extracting the damage symptoms in the plant pest and disease image data and identifying the pest and disease categories in the plant pest and disease image data;

[0037] The emulation evaluation module is responsible for constructing a control state model of the pest and disease category under the standard control effect through a multivariate state estimation method, importing the plant pest and disease image data after the application of botanical pesticides into the control state model to obtain real-time residual data for evaluating the control effect, and obtaining the control effect evaluation results of the botanical pesticide application on the pests and diseases of the plants in the target control area;

[0038] The evaluation result output module is responsible for outputting the control effect evaluation results corresponding to the botanical pesticides and performing visual display.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] This method uses multivariate state estimation to construct a data-driven model, enabling rapid and accurate evaluation of the efficacy of botanical pesticides. It analyzes the effectiveness of different botanical pesticides against target pests and diseases, providing a data foundation for botanical pesticide screening during pest control. Furthermore, based on botanical pesticide efficacy data, it prioritizes pest control options at different stages, facilitating scientific plant pest control. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments or exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or exemplary descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained according to these drawings without paying any creative work.

[0042] Figure 1 A flow chart showing a method for evaluating the pest control effect of botanical pesticides;

[0043] Figure 2 A flowchart of an embodiment showing the construction of a control state model of the current pest and disease category under standard control efficacy by a multivariate state estimation method is shown;

[0044] Figure 3 A flow chart showing an embodiment of evaluating the control effect of botanical pesticides using a control status model;

[0045] Figure 4 A block diagram of a plant-derived pesticide pest control effect evaluation system is shown. DETAILED DESCRIPTION

[0046] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0048] Figure 1 A flow chart showing a method for evaluating the pest control effect of botanical pesticides.

[0049] like Figure 1 As shown, this embodiment provides a method for evaluating the pest control effect of botanical pesticides, comprising:

[0050] S102, acquiring plant disease and insect pest image data within the target control area, pre-processing the collected plant disease and insect pest image data, extracting damage symptoms and identifying the disease and insect pest categories in the plant disease and insect pest image data;

[0051] S104, retrieving control data according to the pest and disease category, and screening the control data that meets the standards based on the control effect, constructing a control state model of the pest and disease category under the standard control effect using a multivariate state estimation method, and using the control state model to obtain a baseline probability distribution under the standard control effect;

[0052] S106, importing the plant disease and insect pest image data after the application of the botanical pesticide into the control state model to obtain real-time residual data, and obtaining the probability distribution of the current control state after the application of the botanical pesticide based on the real-time residual data;

[0053] S108, evaluating the control effect by comparing the probability distribution of the current control status with the baseline probability distribution, and obtaining the control effect evaluation result of the botanical pesticide application on the diseases and pests of the plants in the target control area.

[0054] It should be noted that the plant disease and pest images in the target prevention and control area are obtained through visual equipment, and the obtained plant disease and pest images are adaptively pre-processed according to the lighting environment. The brightness, saturation and contrast of the collected images are adjusted, and the background part around the plants is cropped to improve the quality of the obtained plant disease and pest images.

[0055] The preprocessed plant pest and disease images are imported into a pest and disease recognition network based on the U-Net architecture; the encoder of the pest and disease recognition network is used to extract features from the plant pest and disease images, and the encoder replaces the original structure with the AlexNet convolutional neural network that removes the fully connected layer. The AlexNet convolutional neural network usually includes multiple convolutional layers, pooling layers, ReLU activation functions, and several fully connected layers. The output of each layer becomes the input of the next layer. In the encoder of the pest and disease recognition network, feature maps are extracted through convolutional layers with convolution kernels of different sizes, and L2 regularization is used for weight attenuation after each convolution layer to minimize overfitting. After each layer of convolution, batch normalization is used to improve the performance, stability, and learning speed of the pest and disease recognition network, and the dropout technology is used to randomly discard some neurons during training to improve the generalization performance of the network.

[0056] In addition, in the encoder of the pest and disease recognition network, the original ReLU activation function in the AlexNet convolutional neural network is replaced with the ELU activation function. The ELU activation function will produce negative values, which helps the network guide the bias and weights in the correct direction and generate activation during the gradient calculation process. The feature map generated by the ELU activation function is input into the pooling layer, and the feature map is downsampled in the pooling layer to reduce the dimension of the feature map for further feature extraction; the feature map after the pooling operation is imported into the decoder, and the feature map is restored through upsampling and jump connection to obtain the detailed features in the feature map, and the restored feature map of each layer is subjected to binary classification segmentation to obtain the pest and disease infestation area; the pest and disease infestation area is used as local information to generate recognition prediction branches, and the various recognition prediction branches are fused to realize the integration of global information. The damage symptoms of plant pests and diseases are obtained through the global information, and the Sigmoid layer is used to obtain the probability distribution of the pest and disease category label, and the pest and disease category corresponding to the plant in the target prevention and control area is output.

[0057] It should be noted that, according to the acquired pest and disease categories and the plant categories, retrieval tags are established, and big data methods are used to obtain different types of control data based on the retrieval tags, including chemical control data, physical control data and biological control data. The control data are preliminarily screened according to the current growth stage of the plants in the target control area to obtain control data with consistent growth stages; the disease index of different control methods before and after control is extracted from the preliminarily screened control data, and the preferred disease index can be obtained by obtaining the ratio of the size of the corresponding lesions to the number of plant leaves and the ratio of the number of diseased leaves to the total number of plant leaves in the collected plant pest and disease images, and the ratio of the number of pests in the leaves to the total number of insects in the survey; the disease index is used to obtain the control effect of each control data, and the control effect threshold is obtained according to the growth status of the plant, and the control data with a control effect greater than the control effect threshold is screened, and the plant images after control are extracted from the screened control data to construct a standard control effect image set of plants to characterize the control status of the pest and disease category under the standard control effect.

[0058] Figure 2 The flowchart of the embodiment is shown, which constructs a control state model of the current pest and disease category under the standard control effect by using a multivariate state estimation method.

[0059] According to an embodiment of the present invention, a multivariate state estimation method is used to construct a control state model of the current pest and disease category under standard control efficacy. The control state model is used to obtain a baseline probability distribution under standard control efficacy, specifically:

[0060] S202, extracting training samples based on a standard plant pest control image set, modeling the training samples using a multivariate state estimation method, selecting plant image samples from the standard plant pest control image set using a k-nearest neighbor algorithm, and obtaining k training samples with the closest distance to an observation vector corresponding to the input training sample;

[0061] S204, constructing a memory matrix of the prevention and control status model using the obtained k training samples, calculating an estimated vector of the input training sample using the memory matrix, and training the prevention and control status model by replacing the input training sample;

[0062] S206, obtaining the residual between the observation vector and the estimated vector corresponding to the training sample, calculating the mean and variance of the residual, and obtaining the baseline probability distribution of the pest and disease category under the standard control efficacy based on the residual mean and variance.

[0063] It should be noted that the multivariate state estimation method primarily uses a memory matrix composed of training samples from a set of standard plant control effect images to store and memorize the control state characteristics of that category of plants under the standard control effect. The k-nearest neighbor algorithm is used to dynamically select training samples to construct the memory matrix, fully utilizing the training data and improving the utilization rate of the training data. Furthermore, the constructed memory matrix has higher accuracy. The principle behind using the multivariate state estimation method to establish a control state model for evaluating the control effect of botanical pesticides is that if the control effect of a botanical pesticide is normal after application, the distance and residual between the input data and the control data under the standard control effect used as training data are relatively small. Therefore, the control effect of botanical pesticides can be evaluated by the residual value between the estimated value output by the control state model and the input data.

[0064] During the training process of the prevention and control status model, the obtained residuals are transformed so that the transformed residuals obey the normal distribution. The mean and variance are calculated based on the preprocessed residuals to obtain the baseline probability distribution of the current plant under the corresponding standard prevention effect of this type of pests and diseases.

[0065] Figure 3 A flow chart of an embodiment of evaluating the control effect of botanical pesticides using a control status model is shown.

[0066] According to an embodiment of the present invention, the control effect of botanical pesticides is evaluated using a control state model, specifically:

[0067] S302, obtaining an optimal sampling time after application of the botanical pesticide in the target control area according to the mechanism of action of the botanical pesticide, and obtaining corresponding plant disease and insect pest image data based on the optimal sampling time;

[0068] S304, importing the plant disease and insect pest image data after the application of the botanical pesticide into the control state model, obtaining an observation vector corresponding to the plant disease and insect pest image data after the application of the botanical pesticide, and constructing a memory matrix based on the observation vector;

[0069] S306, calculating an estimated vector using the memory matrix, obtaining real-time residual data output by the control status model based on a comparison between the estimated vector and the observed vector, and performing mean and variance calculations based on the real-time residual data to obtain a probability distribution of the corresponding control status after application of the botanical pesticide;

[0070] S308, obtaining a deviation between the probability distribution of the control state corresponding to the botanical pesticide and the reference probability distribution, reading the maximum component of the deviation, and mapping the maximum component to the interval (0, 1) using a sigmoid function as the deviation degree;

[0071] S310, evaluate the control effect of botanical pesticides based on the deviation, preset the botanical pesticide control effect level and the corresponding numerical range of each level, determine the numerical range into which the deviation falls, read the corresponding control effect level according to the numerical range into which it falls, and output the final evaluation result.

[0072] It should be noted that the plant disease and insect pest image data after the application of botanical pesticides is imported into the control state model to obtain the observation vector X of the plant disease and insect pest image data after the application of botanical pesticides. s , construct a memory matrix D based on the observation vector, and calculate the estimated vector X according to the memory matrix g , by comparing the current estimated vector with the input observation vector to obtain real-time residual data, based on which the probability distribution of the corresponding control state after the application of botanical pesticides is obtained; the calculation formula of the estimated vector is:

[0073]

[0074] Where T represents the matrix transpose, Represents a nonlinear operator, usually a Euclidean distance nonlinear operator.

[0075] It should be noted that when the pest and disease control effect level in the target control area after the application of botanical pesticides is less than the preset level threshold, the disease characteristics of the current pest and disease damage are determined based on the collected plant pest and disease image data; similarity calculation is performed in the screened control data based on the disease characteristics to obtain control data that meets the similarity requirements, and the control measure category corresponding to the control data is determined, and green control measures are screened according to the control measure category; a pest and disease epidemic duration prediction module is constructed based on a multi-layer perceptron, and a corresponding number of prediction heads are set according to the quantity information of the green control measures, and the model is trained according to the control data corresponding to the screened green control measures; the disease characteristics of the current pest and disease damage are used as model input, and the pest and disease epidemic duration in the target control area under different green control measures is obtained, the green control measure with the shortest pest and disease epidemic duration is selected for output, and a comprehensive control plan is generated in combination with botanical pesticides.

[0076] A database is constructed based on the control efficacy data corresponding to different categories of plant-derived pesticides and the disease characteristics of pests and diseases during the prevention and control process. In the database, the corresponding changes in the incidence of pests and diseases are obtained according to the changes in the disease characteristics. The incidence of pests and diseases is obtained by the ratio of the pest and disease plants to the surveyed plants. Based on the changes in the incidence of pests and diseases, the categories of plant-derived pesticides that meet the preset change standards are screened; the plant characteristics and environmental characteristics of the plants in the target area are obtained, and the similarity is calculated based on the plant characteristics and environmental characteristics based on the historical occurrence data of pests and diseases. The incidence of pests and diseases corresponding to each category of pests and diseases in the target area is determined based on the similarity, and the change in the incidence of pests and diseases of each category of pests and diseases is continuously detected and monitored in a preset time step. According to the current change in the incidence of pests and diseases, the category of plant-derived pesticide with the most similarity is retrieved from the database to make recommendations for pest and disease control of plants in the target area.

[0077] Figure 4 A block diagram of a plant-derived pesticide pest control effect evaluation system is shown.

[0078] The second embodiment of the present invention provides a plant-derived pesticide pest control effect evaluation system 4, comprising: a data acquisition module 401, a pest identification module 402, an emulation evaluation module 403 and an evaluation result output module 404;

[0079] The data acquisition module is responsible for acquiring plant disease and insect pest image data within the target control area and preprocessing the acquired plant disease and insect pest image data;

[0080] The pest and disease identification module is responsible for extracting the damage symptoms in the plant pest and disease image data and identifying the pest and disease categories in the plant pest and disease image data;

[0081] The emulation evaluation module is responsible for constructing a control state model of the pest and disease category under the standard control effect through a multivariate state estimation method, importing the plant pest and disease image data after the application of botanical pesticides into the control state model to obtain real-time residual data for evaluating the control effect, and obtaining the control effect evaluation results of the botanical pesticide application on the pests and diseases of the plants in the target control area;

[0082] The evaluation result output module is responsible for outputting the control effect evaluation results corresponding to the botanical pesticides and performing visual display.

[0083] The third aspect of the present invention provides a computer-readable storage medium, which includes a program for evaluating the pest and disease control effect of botanical pesticides. When the program for evaluating the pest and disease control effect of botanical pesticides is executed by a processor, the steps of the method for evaluating the pest and disease control effect of botanical pesticides are implemented.

[0084] In the several embodiments provided in this application, it should be understood that the disclosed methods and systems can be implemented in other ways. The system embodiments described above are merely illustrative. For example, the division of the modules is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple modules or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms. In addition, the functional modules in the various embodiments of the present invention can all be integrated into one processing module, or each module can be a separate module, or two or more modules can be integrated into one module; the above-mentioned integrated modules can be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0085] Those skilled in the art will appreciate that all or part of the steps of the above-mentioned method embodiments may be implemented by hardware associated with program instructions, and the aforementioned program may be stored in a computer-readable storage medium. When the program is executed, the program executes the steps of the above-mentioned method embodiments. The aforementioned storage medium includes various media that can store program codes, such as mobile storage devices, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0086] Alternatively, if the above-mentioned integrated module of the present invention is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0087] The above description is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for evaluating the pest control effect of botanical pesticides, characterized in that: The following steps are involved: Acquire plant disease and insect pest image data within the target control area, pre-process the collected plant disease and insect pest image data, extract damage symptoms and identify the disease and insect pest categories in the plant disease and insect pest image data; Retrieving control data according to the pest and disease category, and screening the control data that meets the standards using the control effect, constructing a control state model of the pest and disease category under the standard control effect using a multivariate state estimation method, and using the control state model to obtain a baseline probability distribution under the standard control effect; Importing plant pest and disease image data after application of botanical pesticides into the control status model to obtain real-time residual data, and obtaining a probability distribution of the current control status after application of the botanical pesticide based on the real-time residual data; By comparing the probability distribution of the current control status with the baseline probability distribution, the control effect is evaluated to obtain the control effect evaluation results of the diseases and pests of plants in the target control area by applying botanical pesticides.

2. The method for evaluating the pest control effect of a botanical pesticide according to claim 1, wherein: Obtain plant disease and insect pest image data within the target control area, pre-process the collected plant disease and insect pest image data, extract damage symptoms and identify the disease and insect pest categories in the plant disease and insect pest image data, specifically: Acquire plant pest and disease images in the target control area through visual equipment, adaptively preprocess the acquired plant pest and disease images according to the lighting environment, and import the preprocessed plant pest and disease images into the pest and disease recognition network based on the U-Net architecture; Using the encoder of the pest and disease recognition network to extract features from plant pest and disease images, the encoder replaces the original structure with an AlexNet convolutional neural network with the fully connected layer removed, extracts feature maps through convolutional layers with convolution kernels of different sizes, and uses L2 regularization for weight decay after each convolution layer to minimize overfitting; The extracted feature map is activated by the ELU activation function and input into the pooling layer, where the feature map is downsampled to reduce the dimension of the feature map for further feature extraction; The feature map after the pooling operation is imported into the decoder, and the feature map is restored through upsampling and skip connection to obtain the detailed features in the feature map. The feature map after each layer is restored is subjected to binary classification segmentation to obtain the pest and disease damage area; The pest and disease infested area is used as local information to generate recognition and prediction branches, and the various recognition and prediction branches are fused to achieve global information integration. The damage symptoms of plant pests and diseases are obtained through the global information, and the Sigmoid layer is used to obtain the probability distribution of the pest and disease category label, and the pest and disease category corresponding to the plant in the target prevention and control area is output.

3. The method for evaluating the pest control effect of a botanical pesticide according to claim 1, wherein: Retrieve control data based on the pest and disease category, and use the control effect to select control data that meets the standards, specifically: Establish search tags based on the acquired pest and disease categories and plant categories, use big data methods to obtain different types of control data based on the search tags, and preliminarily screen the control data based on the current growth stage of the plants in the target control area to obtain control data consistent with the growth stage; The disease index of different control methods before and after control is extracted from the preliminarily screened control data, the control effect of each control data is obtained using the disease index, and a control effect threshold is obtained based on the growth condition of the plant. The control data with a control effect greater than the control effect threshold is screened, and plant images after control are extracted from the screened control data. A standard control effect image set of plants is constructed to characterize the control status of the pest and disease category under the standard control effect.

4. The method for evaluating the pest control effect of a botanical pesticide according to claim 1, wherein: A multivariate state estimation method is used to construct a control state model for the current pest and disease category under standard control efficacy. The control state model is used to obtain the baseline probability distribution under standard control efficacy, specifically: Extracting training samples based on a standard plant pest control effect image set, modeling the training samples using a multivariate state estimation method, selecting plant image samples from the standard plant pest control effect image set using a k-nearest neighbor algorithm, and obtaining k training samples with the closest distance to the observation vector corresponding to the input training sample; Using the obtained k training samples to construct a memory matrix of the prevention and control state model, using the memory matrix to calculate the estimated vector of the input training sample, and training the prevention and control state model by replacing the input training sample; Obtain the residuals of the observation vector and the estimated vector corresponding to the training sample, calculate the mean and variance of the residuals, and obtain the baseline probability distribution of the pest and disease category under the standard control efficacy based on the residual mean and variance.

5. The method for evaluating the pest control effect of a botanical pesticide according to claim 1, wherein: The plant disease and insect pest image data after the application of botanical pesticides is imported into the control state model to obtain real-time residual data. Based on the real-time residual data, the probability distribution of the current control state after the application of botanical pesticides is obtained, specifically: Obtaining an optimal sampling time after applying the botanical pesticide in the target control area according to the mechanism of action of the botanical pesticide, and obtaining corresponding plant disease and insect pest image data based on the optimal sampling time; Importing the plant disease and insect pest image data after the application of the botanical pesticide into the control state model, obtaining an observation vector corresponding to the plant disease and insect pest image data after the application of the botanical pesticide, and constructing a memory matrix based on the observation vector; An estimated vector is calculated using the memory matrix, and real-time residual data output by the control status model is obtained based on the comparison between the estimated vector and the observed vector. The mean and variance of the real-time residual data are calculated to obtain the probability distribution of the corresponding control status after the application of the botanical pesticide.

6. The method for evaluating the pest control effect of a botanical pesticide according to claim 1, wherein: By comparing the probability distribution of the current control status with the baseline probability distribution, the control effect is evaluated to obtain the control effect evaluation results of the plant pests and diseases in the target control area. Specifically: Obtain the deviation value between the probability distribution of the control state corresponding to the botanical pesticide and the baseline probability distribution, read the maximum component of the deviation value, and map the maximum component to the (0, 1) interval using the sigmoid function as the deviation degree; The control effect of botanical pesticides is evaluated based on the deviation degree, the control effect level of botanical pesticides and the corresponding numerical range of each level are preset, the numerical range into which the deviation degree falls is determined, the corresponding control effect level is read according to the numerical range into which it falls, and the final evaluation result is output.

7. The method for evaluating the pest control effect of a botanical pesticide according to claim 6, characterized in that: When the pest control effect level of the target control area after the application of botanical pesticides is less than the preset level threshold, the disease characteristics of the current pest damage are determined based on the collected plant pest image data; Performing similarity calculation on the screened prevention and control data according to the disease characteristics, obtaining prevention and control data that meets the similarity requirements, determining the prevention and control measure categories corresponding to the prevention and control data, and screening green prevention and control measures according to the prevention and control measure categories; A pest and disease epidemic duration prediction module is constructed based on a multi-layer perceptron. A corresponding number of prediction heads are set according to the quantity information of the green control measures. Model training is performed based on the control data corresponding to the selected green control measures. The current disease characteristics of pests and diseases are used as model input to obtain the duration of pest and disease epidemics in the target prevention and control area under different green control measures. The green control measure with the shortest duration of pest and disease epidemics is selected for output, and a comprehensive prevention and control plan is generated in combination with botanical pesticides.

8. A plant-derived pesticide pest control effect evaluation system, characterized in that: A method for evaluating the pest control effect of a botanical pesticide according to any one of claims 1 to 7, comprising: a data acquisition module, a pest identification module, an emulation evaluation module, and an evaluation result output module; The data acquisition module is responsible for acquiring plant disease and insect pest image data within the target control area and preprocessing the acquired plant disease and insect pest image data; The pest and disease identification module is responsible for extracting the damage symptoms in the plant pest and disease image data and identifying the pest and disease categories in the plant pest and disease image data; The emulation evaluation module is responsible for constructing a control state model of the pest and disease category under the standard control effect through a multivariate state estimation method, importing the plant pest and disease image data after the application of botanical pesticides into the control state model to obtain real-time residual data for evaluating the control effect, and obtaining the control effect evaluation results of the botanical pesticide application on the pests and diseases of the plants in the target control area; The evaluation result output module is responsible for outputting the control effect evaluation results corresponding to the botanical pesticides and performing visual display.

Images