Intelligent agricultural management system based on big data

Abstract

The invention discloses an intelligent agricultural management system based on big data. An operation method of the system comprises the following steps of 1, collecting and labeling crop disease images through a network source; 2, establishing a crop disease identification model by using a convolutional neural network; 3, the user uploads a plant disease image in the system and carries out feature description; 4, performing crop disease identification on the image uploaded by the user; 5, the system estimates the probability of each pest type and generates feedback information for the user, and the data acquisition and preprocessing module is used for collecting, describing and preparing training data; the crop disease recognition model establishment module is used for training and optimizing a recognition model; the user interface and disease identification module is used for user verification, image uploading, feature description and real-time disease identification and feedback; the method has the characteristics of accurate, efficient and objective crop disease diagnosis.

Description

Technical Field

[0001] The present invention relates to the field of big data technology, and in particular to an intelligent agricultural management system based on big data. Background Art

[0002] In the existing technology, crop disease problems have always been one of the major challenges in agricultural production. Traditional disease identification methods usually rely on manual observation and expert experience, which is not only time-consuming and labor-intensive, but also easily affected by subjective factors, resulting in inconsistent diagnosis results. Therefore, a faster, more accurate and objective method is needed to solve this problem.

[0003] With the continuous development of big data and computer vision technology, image recognition and deep learning have become effective ways to solve the problem of crop disease identification. Some systems in the existing technology have begun to use deep learning techniques, such as convolutional neural networks (CNNs), to automatically identify crop disease images. However, these systems usually only focus on the image itself and ignore other important information, such as feature descriptions provided by users and environmental factors. In addition, many systems are limited to specific crops and disease types and lack versatility and scalability. Therefore, it is necessary to design an intelligent agricultural management system based on big data that can diagnose crop diseases accurately, efficiently and objectively. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent agricultural management system based on big data to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: an intelligent agricultural management system based on big data, the operation method of the system comprises the following steps: Step 1: Collect and annotate diseased crop images through network sources; Step 2: Use convolutional neural network to establish a crop disease recognition model; Step 3: The user uploads plant disease images in the system and describes the features; Step 4: Identify crop diseases on images uploaded by users; Step 5: The system estimates the probability of each pest type and generates feedback information for the user.

[0006] According to the above technical solution, the step of collecting and labeling diseased crop images through a network source includes: Establish a comprehensive crop disease database, including diverse images, location information, meteorological data, and pest characteristics information; Data tagging and detailed description, including the time the image was taken, location, associated plant type, pest species, and pest severity; Data preprocessing, such as image denoising, data alignment, and format unification, to ensure the quality and consistency of training data.

[0007] According to the above technical solution, the steps of establishing a comprehensive crop disease database, including various images, location information, meteorological data and pest characteristic information, include: Collect diseased crop images on the Internet. These images cover different crop types, different stages of disease development, different lighting conditions or shooting angles, and collect meteorological data of pests related to the shooting location and time of each image, such as temperature, humidity, wind speed and precipitation. In addition, it is also necessary to obtain characteristic data of different pests, such as the life cycle of pests, degree of damage, and host plant information, in order to establish a rich and comprehensive pest database.

[0008] According to the above technical solution, the step of establishing a crop disease recognition model using a convolutional neural network includes: The data was divided into 70% training set, 15% validation set and 15% test set, and then a convolutional neural network architecture was selected for model construction. Specifically, the ResNet-50 model was used, and the pre-trained weight parameters were adopted to accelerate training and improve model performance. In addition, a fully connected layer was added with output neurons matching the number of disease types for multi-category classification. At the same time, data enhancement techniques were applied to increase the diversity of training data, including random rotation, horizontal and vertical flipping, random scaling and random cropping, so as to better generalize the model to different image changes. The model was then trained, and a stochastic gradient descent optimizer with learning rate decay was used to minimize the cross entropy loss. The initial learning rate was set to 0.001 and the batch size was 32. The training process forward propagated the data, calculated the loss, performed backpropagation and updated the model weights to continuously reduce the error, and set an early stopping strategy. If the validation loss did not improve in five consecutive rounds, the training would be terminated. The training process will stop and use the validation set data to evaluate the performance of the model, measuring the accuracy, recall, F1 score and confusion matrix performance indicators. The performance on the validation set determines whether the system needs to further adjust the model or change the hyperparameters. If the performance is not up to standard, the system will perform hyperparameter adjustments, including adjustment of the learning rate, selection of the optimizer and fine-tuning of the model architecture. In addition, the system uses learning rate decay. The initial learning rate is set to 0.001 and is reduced to half of the original every 10 rounds. In order to reduce the overfitting of the model, a Dropout layer is also used, and the dropout rate is set to 0.5 to randomly turn off half of the neurons during training. Once the model is trained, it is deployed on a cloud server for real-time disease recognition. Users can use mobile applications to take images and upload them to obtain disease recognition results. Developers regularly monitor model performance and retrain when necessary to adapt to new data and disease types, which ensures that the model maintains high performance in a changing environment.

[0009] According to the above technical solution, the step of the user uploading a plant disease image in the system and performing feature description includes: Users need to provide a username and password or create an account through other verification methods. After successful login, the application will actively obtain the location information of the device. Subsequently, users can use the application to upload the captured plant disease images. Image processing functions will be provided to perform compression and format conversion before uploading to reduce the size of data transmission. After the user uploads the image, the application will provide a feature description interface. The user can provide detailed feature descriptions of the plant disease, such as the appearance of the symptoms, the affected parts, and the possible degree of disease. These descriptions will supplement the image information and help the system identify the problem more accurately. The application will communicate with the crop disease identification system in real time to obtain instant identification feedback. Through this step, users can be provided with a convenient tool to enable them to upload plant disease images, provide feature descriptions, and obtain timely disease identification feedback. Such a user interface and application will help farmers better manage their crops and take timely measures to deal with plant disease problems.

[0010] According to the above technical solution, the step of identifying crop diseases on the images uploaded by the user includes: The system extracts keywords added by the user in the feature information, which includes the user's description of the disease characteristics, such as the appearance of the leaves, the affected parts, and the possible degree of disease. Then, using natural language processing technology, the system will match the extracted keywords. After matching, the system will integrate the image recognition results and the feature information provided by the user. The system will also obtain the geographic information of the user's current location. Then, the system will connect to the meteorological data source to obtain the meteorological data of the user's current location, including temperature, humidity, and precipitation. This data will be used to analyze the current environmental conditions to predict the types of pests that may occur. Combining the image recognition results, the feature information provided by the user, the location information, and the meteorological data, the system will predict the types of pests that may occur. This step will provide more comprehensive information to help users understand potential disease risks.

[0011] According to the above technical solution, the system estimates the probability of each pest type and generates feedback information for the user, including: Consider the prior probability of pest type; Calculate conditional probabilities, including image features, keywords, temperature and humidity data; The posterior probability of each pest type was calculated using Bayes’ theorem; The pest type with the highest posterior probability is selected as the prediction result; Provides user feedback to help them understand potential pest problems and take appropriate actions.

[0012] According to the above technical solution, the step of selecting the pest type with the highest posterior probability as the prediction result includes: Consider the prior probability of the pest type (P(pest)), which represents the probability of each pest type occurring in the absence of other information. This can be estimated based on the pest types prevalent in the current season. Next, focus on the conditional probabilities, which include image features, keywords, temperature, and humidity data. Then calculate the conditional probabilities, P(image features|pests), P(keywords|pests), P(temperature|pests), and P(humidity|pests). These conditional probabilities represent the probability of matching the relevant information given a certain pest type. Finally, consider the marginal probabilities of the features, namely P(image features), P(keywords), P(temperature), and P(humidity). These are the probabilities of occurrence of the respective features regardless of the pest type. By using these probability values, the system can use Bayes' theorem to calculate the posterior probability of each pest type (P(pests|image features, keywords, temperature, humidity)). The pest type with the highest posterior probability will be considered the most likely pest type.

[0013] According to the above technical solution, the system includes: Data acquisition and preprocessing module, used to collect, describe and prepare training data; Crop disease identification model building module, used to train and optimize the identification model; User interface and disease identification module for user authentication, image upload, feature description, and real-time disease identification and feedback.

[0014] According to the above technical solution, the data acquisition and preprocessing module includes: The data collection module is used to collect diseased crop images from the Internet, including images of various crop types, different disease stages, and under different conditions, as well as relevant meteorological data and pest characteristic data; Data tagging and description module, used to tag and describe the collected images and data in detail, including shooting information, plant type, and pest type; The data preprocessing module is used to preprocess the collected data, including image denoising, data alignment and format unification to ensure the quality and consistency of the training data.

[0015] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: In the present invention, the system first collects diseased crop images on the Internet and establishes a comprehensive disease database, which covers various crop types and pest conditions to ensure data diversity. Then, the data is marked, cleaned and preprocessed to ensure high quality and consistency. Subsequently, a crop disease recognition model is established through a convolutional neural network, and trained using a ResNet-50 model and data enhancement technology, considering hyperparameter adjustment and performance evaluation. Next, a user-friendly mobile application is provided, allowing users to upload plant disease images and provide feature descriptions, while obtaining location information. Step four uses the trained model to recognize images uploaded by users, extract feature information, and estimate the probability of pest types, considering image recognition results, keyword matching and meteorological data. Finally, feedback information is generated to help users understand potential pest risks and provide them with disease identification and suggestions. The method has the characteristics of accurate, efficient and objective crop disease diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 A flowchart of a smart agriculture management method based on big data provided in Embodiment 1 of the present invention; Figure 2 A schematic diagram of the module composition of the big data-based intelligent agricultural management system provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0017] 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.

[0018] Embodiment 1: Figure 1 This is a flow chart of a smart agricultural management method based on big data provided in Embodiment 1 of the present invention. This embodiment can be applied to the scenario of crop disease identification. This method can be executed by the smart agricultural management system based on big data provided in this embodiment. Figure 1 As shown, the method specifically comprises the following steps: Step 1: Collect and annotate diseased crop images through network sources; In an embodiment of the present invention, a comprehensive crop disease database is established to support subsequent big data training and accurate crop disease identification; For example, diseased crop images on the Internet are collected. These images cover different crop types, different disease development stages, different lighting conditions or shooting angles, and meteorological data of pests related to the shooting location and time of each image, such as temperature, humidity, wind speed and precipitation, etc. In addition, characteristic data of different pests, such as detailed information such as the life cycle of pests, degree of damage, host plants, etc., are also required to establish a rich and comprehensive pest database; Exemplarily, each piece of collected data needs to be labeled and described in detail. For images of diseased crops collected on the Internet, it is necessary to describe in detail information such as the shooting time, location, related plant type, type of pest, and severity of the pest. The collected data is then uploaded to a cloud storage platform, and indexes and metadata are established for the data, and permission control is set. After the data is collected, data cleaning and preprocessing are performed, such as image denoising, data alignment, and data format unification, to ensure the quality and consistency of the training data. Through this step, a crop disease data set can be established, which provides a solid foundation for subsequent big data training and accurate crop disease identification.

[0019] Step 2: Use convolutional neural network to establish a crop disease recognition model; In the embodiment of the present invention, a convolutional neural network is used to train the collected data, thereby establishing a crop disease recognition model; Exemplarily, the data is divided into 70% training set, 15% validation set and 15% test set, and then a convolutional neural network architecture is selected for model construction. Specifically, a ResNet-50 model is used, and pre-trained weight parameters are adopted to accelerate training and improve model performance. In addition, a fully connected layer is added, which has output neurons matching the number of disease types for multi-category classification. At the same time, data enhancement techniques are applied to increase the diversity of training data, including random rotation, horizontal and vertical flipping, random scaling and random cropping, so as to better generalize the model to different image changes. Then, the model is trained, and a stochastic gradient descent optimizer with learning rate decay is used to minimize the cross entropy loss. The initial learning rate is set to 0.001, and the batch size is 32. The training process forward propagates the data, calculates the loss, performs backpropagation and updates the model weights to continuously reduce the error, and sets an early stopping strategy. If the verification loss does not improve in 5 consecutive rounds, the training will stop; Exemplarily, the validation set data is used to evaluate the performance of the model, and performance indicators such as accuracy, recall, F1 score and confusion matrix are measured. The performance on the validation set determines whether the system needs to further adjust the model or change hyperparameters. If the performance is not up to standard, the system will perform hyperparameter adjustments, including adjustment of the learning rate, selection of the optimizer and fine-tuning of the model architecture. In addition, the system uses learning rate decay. The initial learning rate is set to 0.001 and is reduced to half of the original every 10 rounds. In order to reduce the overfitting of the model, a Dropout layer is also used, and the dropout rate is set to 0.5 to randomly turn off half of the neurons during training. Once the model is trained, it is deployed on a cloud server for real-time disease recognition. Users can use mobile applications to take images and upload them to obtain disease recognition results. Developers regularly monitor model performance and retrain when necessary to adapt to new data and disease types, which ensures that the model maintains high performance in a changing environment.

[0020] Step 3: The user uploads plant disease images in the system and describes the features; In an embodiment of the present invention, through the user interface, users can conveniently upload plant disease images, provide feature descriptions, and obtain immediate recognition feedback; Exemplarily, the user needs to provide a username and password or create an account through other verification methods. After successful login, the application will actively obtain the location information of the device. Subsequently, the user can use the application to upload the captured plant disease images. Image processing functions will be provided to perform compression and format conversion before uploading to reduce the size of data transmission. After the user uploads the image, the application will provide a feature description interface. The user can provide detailed feature descriptions of the plant disease, such as the appearance of the symptoms, the affected parts, and the possible extent of the disease. These descriptions will supplement the image information and help the system to more accurately identify the problem. The application will communicate with the crop disease identification system in real time to obtain instant identification feedback. Through this step, a convenient tool can be provided to users to enable them to upload plant disease images, provide feature descriptions, and obtain timely disease identification feedback. Such a user interface and application will help farmers better manage their crops and take timely measures to deal with plant disease problems.

[0021] Step 4: Identify crop diseases on images uploaded by users; In the embodiment of the present invention, the system uses a pre-trained crop disease recognition model to analyze the plant disease images uploaded by the user, and also receives the user's location information. The model will identify the disease type in the image and provide an initial disease recognition result, laying the foundation for subsequent steps; Exemplarily, the system extracts keywords supplemented by the user in the feature information, which includes the user's description of the disease characteristics, such as the appearance of the leaves, the affected parts, and the possible degree of disease. Then, using natural language processing technology, the system will match the extracted keywords. After matching, the system will integrate the image recognition results and the feature information provided by the user. The system will also obtain the geographic information of the user's current location. Then, the system will connect to the meteorological data source to obtain meteorological data at the user's current location, including temperature, humidity, precipitation, etc. This data will be used to analyze the current environmental conditions to predict the types of pests that may occur. Combining the image recognition results, the feature information provided by the user, the location information and the meteorological data, the system will predict the types of pests that may occur. This step will provide more comprehensive information to help users understand potential disease risks.

[0022] Step 5: The system estimates the probability of each pest type and generates feedback information for the user.

[0023] In an embodiment of the present invention, by combining image recognition results, keyword matching, and meteorological data, the system can estimate the possible pest type; Exemplarily, first, consider the prior probability of the pest type (P(pest)), which represents the probability of each pest type occurring in the absence of other information, which can be estimated based on the pest types prevalent in the current season. Next, focus on the conditional probability, which includes image features, keywords, temperature and humidity data, and then calculate the conditional probabilities, P(image features|pests), P(keywords|pests), P(temperature|pests) and P(humidity|pests). These conditional probabilities represent the probability of matching the relevant information given a certain pest type. Finally, consider the marginal probabilities of the features, namely P(image features), P(keywords), P(temperature) and P(humidity). These are the occurrence probabilities of the respective features without considering the pest type. By using these probability values, the system can use Bayes' theorem to calculate the posterior probability of each pest type (P(pests|image features, keywords, temperature, humidity)). The pest type with the highest posterior probability will be regarded as the most likely pest type. For example, assume there are three types of pests: A, B, C, and the corresponding conditional probabilities and prior probabilities: 1. Prior probability: P(A) = 0.4, P(B) = 0.3, P(C) = 0.3; 2. Conditional probability (P(image features|pests)): P(image feature|A) = 0.8, P(image feature|B) = 0.2, P(image feature|C) = 0.1; 3. Conditional probability (P(keyword|pest)): P(keyword|A) = 0.6, P(keyword|B) = 0.2, P(keyword|C) = 0.1; 4. Conditional probability (P(temperature|pest)) and P(humidity|pest)): P(temperature|A) = 0.7, P(humidity|A) = 0.4; P(temperature|B) = 0.9, P(humidity|B) = 0.3; P(temperature|C) = 0.6, P(humidity|C) = 0.2; 5. Use Bayes’ theorem to calculate the posterior probability (P(pest|image features, keywords, temperature, humidity)): P(A|data) = P(A) * P(image feature|A) * P(keyword|A) * P(temperature|A) * P(humidity|A) / (P(data)) P(B|data) = P(B) * P(image feature|B) * P(keyword|B) * P(temperature|B) * P(humidity|B) / (P(data)) P(C|data) = P(C) * P(image feature|C) * P(keyword|C) * P(temperature|C) * P(humidity|C) / (P(data)) where P(data) is a normalization factor that ensures that the sum of all posterior probabilities equals 1. Finally, by comparing the posterior probabilities, the pest type with the highest posterior probability is selected and used as the system's prediction result. This prediction result will be fed back to users by the system to help them better understand possible pest problems and take corresponding agricultural measures.

[0024] Embodiment 2: Embodiment 2 of the present invention provides an intelligent agricultural management system based on big data. Figure 2 The schematic diagram of the module composition of the smart agricultural management system based on big data provided in the second embodiment of the present invention is as follows: Figure 2 As shown, the system includes: Data acquisition and preprocessing module, used to collect, describe and prepare training data; Crop disease identification model building module, used to train and optimize the identification model; User interface and disease identification module for user authentication, image upload, feature description, and real-time disease identification and feedback; In some embodiments of the present invention, the data acquisition and preprocessing module includes: The data collection module is used to collect diseased crop images from the Internet, including images of various crop types, different disease stages, and under different conditions, as well as relevant meteorological data and pest characteristic data; Data tagging and description module, used to tag and describe the collected images and data in detail, including shooting information, plant type, pest type, etc.; Data preprocessing module, used to preprocess the collected data, including image denoising, data alignment and format unification, to ensure the quality and consistency of training data; In some embodiments of the present invention, the crop disease identification model building module includes: Data segmentation and model selection module, which is used to divide the data into training set, validation set and test set, and select the convolutional neural network architecture (ResNet-50) for model building; Model training and optimization module, which is used to train the model using stochastic gradient descent optimizer, learning rate decay, batch size and other settings; Model deployment module, used to deploy the trained model on the cloud server to provide users with real-time disease identification services; In some embodiments of the present invention, the user interface and disease identification module include: User authentication and data upload module, used by users to create accounts or verify identities through the application and upload captured plant disease images; A feature description module is used to provide a user interface for users to provide detailed feature descriptions of plant diseases, such as symptoms, affected parts, and disease severity; The image recognition and feedback module is used to use the trained model to identify crop diseases in images uploaded by users.

[0025] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article 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, method, article or device.

[0026] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A smart agricultural management method based on big data, the method comprising the following steps: Step 1: Collect and annotate diseased crop images through network sources; Step 2: Use convolutional neural network to establish a crop disease recognition model; Step 3: The user uploads plant disease images in the system and describes the features; Step 4: Identify crop diseases on images uploaded by users; Step 5: The system estimates the probability of each pest type and generates feedback information for the user; The step of collecting and labeling diseased crop images through a network source includes: Establish a comprehensive crop disease database, including diverse images, location information, meteorological data, and pest characteristics information; Data tagging and detailed description, including the time the image was taken, location, associated plant type, pest species, and pest severity; Data preprocessing, such as image denoising, data alignment, and format unification, to ensure the quality and consistency of training data; The system estimates the probability of each pest type and generates feedback information for the user, comprising: Consider the prior probability of pest type; Calculate conditional probabilities, including image features, keywords, temperature and humidity data; The posterior probability of each pest type was calculated using Bayes’ theorem; The pest type with the highest posterior probability is selected as the prediction result; Provide user feedback to help them understand potential pest problems and take appropriate measures; The step of selecting the pest type with the highest posterior probability as the prediction result comprises: Consider the prior probability of the pest type, P(pest), which represents the probability of each pest type occurring in the absence of other information. This can be estimated based on the pest types prevalent in the current season. Next, focus on the conditional probabilities, which include image features, keywords, temperature, and humidity data. Then calculate the conditional probabilities, P(image features|pests), P(keywords|pests), P(temperature|pests), and P(humidity|pests). These conditional probabilities represent the probability of matching the relevant information given a certain pest type. Finally, consider the marginal probabilities of the features, namely P(image features), P(keywords), P(temperature), and P(humidity). These are the probabilities of occurrence of the respective features regardless of the pest type. By using these probability values, the system can use Bayes' theorem to calculate the posterior probability of each pest type, P(pests|image features, keywords, temperature, humidity). The pest type with the highest posterior probability will be considered the most likely pest type.

2. Intelligent agricultural management system based on big data, characterized by: The smart agriculture management system implements the smart agriculture management method based on big data according to claim 1, and the system comprises: Data acquisition and preprocessing module, used to collect, describe and prepare training data; Crop disease identification model building module, used to train and optimize the identification model; User interface and disease identification module for user authentication, image upload, feature description, and real-time disease identification and feedback.

3. The intelligent agricultural management system based on big data according to claim 2 is characterized in that: The data acquisition and preprocessing module includes: The data collection module is used to collect diseased crop images from the Internet, including images of various crop types, different disease stages, and under different conditions, as well as relevant meteorological data and pest characteristic data; Data tagging and description module, used to tag and describe the collected images and data in detail, including shooting information, plant type, and pest type; The data preprocessing module is used to preprocess the collected data, including image denoising, data alignment and format unification to ensure the quality and consistency of the training data.

Images