An intelligent detection system for wheat pests and diseases

Abstract

The present invention discloses an intelligent detection system for wheat diseases and insect pests, which relates to the technical field of disease and insect pest detection, including a collection end, an application server end and a client end, wherein the collection end is used to collect wheat leaf photos and soil samples of wheat fields, the application server end is used to detect the collected soil samples and analyze the leaf photos, and then used to determine the pathogens of wheat leaf diseases and insect pests, and the client end is used to display the detection results and specific prevention and control measures, so as to facilitate users to manage wheat fields. The intelligent detection system for wheat diseases and insect pests analyzes leaf photos based on the target detection model of the lightweight network ShuffleNetv2‑YOLO, can more effectively extract wheat leaf diseases and insect pests characteristics, improve detection accuracy, overcome the limitations of a single detection method through the application of multi-source data fusion and advanced model algorithms, and greatly improve the accuracy and reliability of disease and insect pest detection and diagnosis.

Description

Technical Field

[0001] The invention relates to the technical field of pest and disease detection, and in particular to an intelligent detection system for wheat pest and disease. Background Art

[0002] Wheat fruit is one of the staple foods of human beings. After the mature wheat grains are ground into flour, they can be made into bread, steamed buns, biscuits, noodles and other foods. After fermentation, they can be made into beer, alcohol, liquor (such as vodka), or biomass fuel. It can be seen that wheat occupies an important position in human daily life. Therefore, wheat planting is also a major crop in my country. With the continuous increase in population, increasing wheat production has also become a trend. Only when wheat can grow normally can the stability of autumn production be guaranteed. Therefore, in addition to daily fertilization management, it is also necessary to timely detect whether wheat has insect pests and provide timely treatment for wheat with insect pests.

[0003] The invention patent with announcement number CN118587475A discloses a detection method and processing system suitable for crop pests, which includes the following steps: (1) using the YOLO V5 algorithm to detect targets on the collected crop images and determine the type and growth stage of the pests; (2) using the DeepSort and LSTM algorithms to track the movement trajectory of the pests, divide their distribution areas, and predict their possible expansion locations; (3) using the above collected information to determine the pest control strategy; (4) controlling the agricultural robot to reach the target area and perform corresponding pest control operations.

[0004] Existing detection methods and processing systems generally combine YOLO V5, DeepSort and LSTM algorithms in deep learning to detect possible pests and diseases in agricultural production, and make appropriate pest control strategies based on specific circumstances and distribution areas. However, the model of this method is greatly affected by the environment, has many identification steps and a long time, and is difficult to deploy on mobile terminals for real-time pest and disease detection. It is also subject to the limited storage of databases, making it difficult to promote on a large scale. Summary of the invention

[0005] In view of the deficiencies of the prior art, the present invention provides an intelligent detection system for wheat diseases and insect pests, which solves the existing problems.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent detection system for wheat pests and diseases, comprising:

[0007] A collection end, which is used to collect photos of wheat leaves and soil samples from wheat fields;

[0008] An application server side, which is used to detect the collected soil samples and analyze the leaf photos, and then to determine the pathogens of wheat leaf diseases and insect pests;

[0009] The client is used to display the detection results and specific prevention and control measures, thereby facilitating users to manage wheat fields.

[0010] Preferably, the collection end includes a drone, a PCR detection instrument and a wireless transmission module integrated into the two respectively. The drone takes multi-angle photos of wheat leaves through the camera thereon and uploads the photos to the application server through the wireless transmission module. The PCR detection instrument is used to detect the types and quantities of pathogens in the collected soil samples, and uploads the data to the application server through the wireless transmission module.

[0011] Preferably, a soil sampler is installed inside the drone, a sample tube is installed inside the soil sampler, the soil sampler is used to collect soil samples, and the sample tube is used to automatically package the collected soil samples.

[0012] Preferably, the application server includes a data processing unit, a data storage module, a communication module and a collection unit, and the data processing unit is used to analyze the soil samples collected by the detection and the leaf photos;

[0013] The data storage module is used to store data obtained from the Internet and uploaded by users;

[0014] The communication module is used to realize the communication connection between the acquisition end, the client end and the Internet by using the server end;

[0015] The collection unit is used to collect wheat-related pest and disease image data, wheat pest and disease control measure data, and relevant medicines for controlling wheat pest and disease on the Internet.

[0016] Preferably, the application server also includes a model creation unit, which improves the target detection algorithm model of YOLO based on the lightweight network ShuffleNetv2 and flexible non-maximum suppression Soft-NMS to construct a new target detection network model ShuffleNetv2-YOLO.

[0017] Preferably, the specific structure of the target detection network model ShuffleNetv2-YOLO is: replacing the feature extraction backbone network Backbone in the YOLO model with the CBRM module in the lightweight network ShuffleNetv2, and replacing the non-maximum suppression function NMS in YOLO with the flexible non-maximum suppression function Soft-NMS.

[0018] Preferably, the data processing unit includes a blade photo data set construction module, a model calculation and comparison module, a judgment and analysis module, and a measure generation module, and each module is connected through signals;

[0019] The leaf photo dataset construction module is used to construct a rice leaf photo dataset;

[0020] The model calculation comparison module is used to input the leaf photo dataset into the target detection network model for comparison calculation;

[0021] The judgment and analysis module is used to judge and analyze the results output by the model and the data uploaded by the PCR detection instrument;

[0022] The measure generation module provides prevention and control measures for the diseases and insect pests that appear in wheat based on the analysis results of the judgment and analysis module.

[0023] Preferably, the leaf photo dataset construction module is also used to adjust the image to a square fixed pixel of 640*640, and normalize the pixel value of the collected image to between 0 and 1, so as to reduce the image inconsistency caused by lighting changes or the influence of other external factors.

[0024] Preferably, the client includes a display module for displaying and explaining the data statistical information in the resource library. The display and explanation of the data statistical information refers to selecting bar charts, line charts, word clouds, heat maps, etc. according to data types and characteristics to display and explain the number of pictures, number of news, number of videos, categories and numbers of pest and disease pictures, types and numbers of news, sources of news, and word clouds of news keywords in the database.

[0025] Beneficial Effects

[0026] The present invention provides an intelligent detection system for wheat pests and diseases. Compared with the prior art, it has the following beneficial effects:

[0027] 1. The intelligent detection system for wheat pests and diseases can comprehensively and quickly collect wheat leaf photos and soil samples through drones equipped with cameras and soil samplers, and realize the acquisition of large-scale wheat field data. The target detection model based on the lightweight network ShuffleNetv2-YOLO analyzes the leaf photos. Its unique network structure design, such as using the CBRM module in ShuffleNetv2 to replace the traditional feature extraction backbone network, can more effectively extract the characteristics of wheat leaf pests and diseases and improve the detection accuracy. Combined with the detection data of soil pathogens by PCR detection instruments, the comprehensive judgment and analysis module can accurately determine the pathogens, types and severity of wheat pests and diseases. Through the application of multi-source data fusion and advanced model algorithms, it overcomes the limitations of a single detection method, greatly improves the accuracy and reliability of pest and disease detection and diagnosis, and provides a key basis for timely and effective prevention and control.

[0028] 2. The intelligent detection system for wheat pests and diseases. The client of the system provides users with an intuitive and convenient management interface, which can clearly display the detection results and prevention and control measures, so that users can quickly understand the pest and disease status of wheat fields and take corresponding actions. The display module presents data statistical information in a variety of visual ways to help users deeply analyze pest and disease related data, such as pest and disease image categories and quantity change trends, key information in news information, etc., which helps users grasp the development trend of pests and diseases and industry dynamics and make more scientific management decisions. The data storage module and collection unit on the application server side integrate various data from the Internet, user uploads and self-detection, including pest and disease images, prevention and control measures, related drugs and other information. It not only enriches the system's data resources and provides more comprehensive data support for model training and analysis, but also forms an agricultural knowledge sharing platform for users to query and refer to at any time, improve the overall intelligence level and efficiency of agricultural production management, and promote sustainable agricultural development. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic diagram of system connection of the present invention;

[0030] Figure 2 It is a structural schematic diagram of a data processing unit of the present invention;

[0031] Figure 3 It is a schematic diagram of the method flow of the present invention; DETAILED DESCRIPTION

[0032] 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.

[0033] See also Figure 1-3 , the present invention provides two technical solutions:

[0034] The first implementation mode: an intelligent detection system for wheat pests and diseases, comprising:

[0035] The collection end is used to collect photos of wheat leaves and soil samples from wheat fields.

[0036] The collection end includes a drone, a PCR detection instrument and a wireless transmission module integrated in the two. The drone uses its camera to take multi-angle photos of wheat leaves and uploads them to the application server through the wireless transmission module. The PCR detection instrument is used to detect the types and quantities of pathogens in the collected soil samples, and uploads the data to the application server through the wireless transmission module.

[0037] The PCR detection instrument uses advanced microfluidic chip technology, which can achieve rapid amplification and detection of pathogens in soil samples on a tiny chip. This technology greatly shortens the detection time from several hours in traditional PCR detection to tens of minutes, while improving the sensitivity and accuracy of the detection. An intelligent sample processing system is integrated into the instrument, which can automatically pre-process the collected soil samples, including sample grinding, dilution, and reagent addition, reducing the errors and cumbersome processes caused by manual operation. Its wireless transmission module can not only upload the test data, but also provide real-time feedback on the instrument's operating status, test progress and other information to the application server for remote monitoring and troubleshooting. In addition, the PCR detection instrument is also equipped with a large-capacity sample storage bin that can store multiple soil samples at the same time to meet the detection needs of large areas of wheat fields. The sample storage bin has a refrigeration function that can maintain the biological activity of the sample for a certain period of time for review or further analysis.

[0038] The drone is equipped with a soil sampler inside, and a sample tube inside the soil sampler. The soil sampler is used to collect soil samples, and the sample tube is used to automatically seal the collected soil samples. The soil sampler is designed with a high degree of automation and precision. It is equipped with a high-precision soil depth sensor, which can accurately collect soil samples according to the preset depth requirements to avoid affecting the test results due to inconsistent collection depths. The sample tube uses special sealing materials to ensure that the collected soil samples are not disturbed by the external environment during transportation and storage, and maintain their original pathogen type and quantity information. The soil sampler also has a self-cleaning function. After each soil collection is completed, it can automatically remove the residual soil inside to prevent cross-contamination and prepare for the next sampling.

[0039] In addition to conventional cameras, the drone is also equipped with a multispectral imager. The multispectral imager can capture the reflection information of wheat in different spectral bands, so as to more accurately analyze the health of wheat leaves. For example, it can detect the chlorophyll content and moisture content of leaves through specific spectral characteristics. These indicators are closely related to whether wheat is suffering from pests and diseases and what kind of pests and diseases it suffers. The flight path planning of the drone is controlled by an intelligent navigation system, which can automatically generate the optimal flight route according to the area, shape and preset sampling density of the wheat field, ensuring full coverage of the entire field and efficient collection of leaf photos. At the same time, the battery system of the drone uses a high-energy-density lithium battery, and has fast charging and automatic return charging functions to ensure long-term and uninterrupted operation needs. Its wireless transmission module supports 5G communication technology, which can achieve high-speed and stable data upload, and can ensure that image data and soil collection information are transmitted to the application server in a timely manner even at a long working distance.

[0040] The application server side is used to detect the collected soil samples and analyze the leaf photos, and then to determine the pathogens of wheat leaf diseases and pests.

[0041] The application server includes a data processing unit, a data storage module, a communication module and a collection unit. The data processing unit is used to analyze the soil samples collected by the detection and the leaf photos;

[0042] The data storage module adopts a layered storage architecture when storing data obtained from the Internet and uploaded by users. Frequently used data (such as recent detection data, frequently accessed pest and disease images, etc.) are stored in the cache layer to increase the data reading speed; historical data and infrequently used data are stored in the large-capacity disk storage layer to reduce storage costs. At the same time, the data storage module implements automatic data backup and redundant storage functions to ensure data security and reliability. Regularly clean and organize the stored data, delete redundant data and expired data, optimize the data storage structure, and improve data query and management efficiency. In addition, the data storage module also has a data encryption function to encrypt and store sensitive data (such as user information, pathogen gene sequences in pest and disease detection data, etc.) to prevent data leakage.

[0043] The communication module uses a fusion of multiple communication protocols and technologies to achieve communication connections between the application server and the collection end, client, and the Internet. In addition to supporting the traditional TCP / IP protocol, the UDP protocol is also optimized for the data transmission needs of drones to improve the real-time and efficiency of data transmission. For communication with the client, the communication module supports a variety of terminal device types, including mobile phones, tablets, computers, etc., and has developed corresponding client application programming interfaces (APIs) for the operating systems of different devices (such as Android, iOS, Windows, etc.) to ensure that the client can communicate with the server stably and conveniently. In terms of connecting to the Internet, the communication module has a firewall and intrusion detection system, which can effectively prevent network attacks and ensure the security of servers and data. At the same time, the communication module also realizes connection with agricultural Internet of Things devices, such as communicating with meteorological sensors, soil moisture sensors and other devices in the field, to obtain more environmental data and provide richer information for the comprehensive analysis of wheat diseases and pests.

[0044] The application server also includes a model creation unit, which improves the target detection algorithm model of YOLO based on the lightweight network ShuffleNetv2 and flexible non-maximum suppression Soft-NMS, and constructs a new target detection network model ShuffleNetv2-YOLO.

[0045] Specifically, the specific structure of the target detection network model ShuffleNetv2-YOLO is: replace the feature extraction backbone network Backbone in the YOLO model with the CBRM module in the lightweight network ShuffleNetv2, and replace the non-maximum suppression function NMS in YOLO with the flexible non-maximum suppression function Soft-NMS.

[0046] The model creation unit conducted in-depth model optimization and parameter adjustment when improving the target detection algorithm model of YOLO based on the lightweight network ShuffleNetv2 and flexible non-maximum suppression Soft-NMS to build a new target detection network model ShuffleNetv2-YOLO. The CBRM module in ShuffleNetv2 was adjusted in parameters and fine-tuned in structure, such as optimizing the size and step size combination of the convolution kernel to better meet the extraction requirements of wheat leaf pest and disease characteristics. When introducing Soft-NMS, a large number of experiments and optimizations were conducted on the parameters of its attenuation function to find the most suitable parameter settings for wheat pest and disease detection scenarios and improve the accuracy and recall rate of target detection. The model creation unit also established a model training monitoring system to monitor the changes in the model's loss function value, accuracy, recall rate and other indicators in real time during the model training process, and timely adjust the training strategy according to the changing trends of these indicators, such as adjusting the learning rate and increasing the amount of training data, to ensure that the constructed ShuffleNetv2-YOLO model has good performance.

[0047] In this embodiment, the application process of the ShuffleNetv2 module in the feature extraction backbone network is:

[0048] Assume that the input feature map X∈R H×W×C (where H is height, W is width, and C is the number of channels), convolution kernel size k, step size s, padding p, and output channel number C out ;

[0049] Convolution operation:

[0050]

[0051] In the formula, is the convolution kernel weight, c is the input channel index, where

[0052] Batch Normalization:

[0053]

[0054] In the formula, μ and σ 2 are the batch mean and variance, γ and β are learnable parameters, ∈ is a small constant to prevent the denominator from being zero,

[0055] In this embodiment, the application process of the flexible non-maximum suppression function Soft-NMS in the non-maximum suppression function NMS is as follows:

[0056] Assume that the initial detection box set D = {d1, d2, ..., d m}, corresponding to the score set S = {s1,s2,…,s m};

[0057] For each detection box d i , calculate its and other boxes d j The overlap IOU (d i ,d j );

[0058] Score Adjustments:

[0059]

[0060] The above formula uses the Gaussian attenuation function, σ is the attenuation parameter, s j is a flexible non-maximum value;

[0061] Finally, the detection frames are screened according to the adjusted scores, and the detection frames with scores higher than the threshold are retained as the final detection results.

[0062] More specifically, the data processing unit includes a blade photo data set construction module, a model calculation and comparison module, a judgment and analysis module, and a measure generation module, and each module is connected through signals;

[0063] The leaf photo dataset construction module is used to construct a rice leaf photo dataset. When constructing the rice leaf photo dataset, in addition to basic image adjustments (such as adjusting to a 640*640 square fixed pixel and normalizing the pixel value), the leaf photo dataset construction module also uses image enhancement technology. For example, random cropping, flipping, rotation and other operations are used to increase the diversity of the dataset and improve the generalization ability of the model. At the same time, the image filtering algorithm is used to remove noise interference in the image to further improve the image quality. The module also establishes connections with multiple agricultural image databases on the Internet, regularly and automatically downloads and updates image data related to wheat pests and diseases, enriches the content of the dataset, and enables it to cover more types of wheat pests and diseases and wheat leaf characteristics under different growth environments.

[0064] The model calculation comparison module adopts a distributed computing architecture when inputting the leaf photo dataset into the target detection network model ShuffleNetv2-YOLO for comparison calculation. Through parallel calculation on multiple high-performance servers, the calculation time is greatly shortened and the detection efficiency is improved. During the calculation process, the model automatically records the intermediate results and related parameters of each calculation for subsequent model optimization and performance analysis. At the same time, the module also implements the automatic evaluation function of the model, and evaluates the output results of the model in real time according to the preset evaluation indicators (such as accuracy, recall rate, F1 value, etc.), and automatically adjusts some hyperparameters of the model according to the evaluation results, such as learning rate, number of iterations, etc., to achieve adaptive optimization of the model.

[0065] The judgment and analysis module introduces an expert knowledge base system when judging and analyzing the results of the model output and the data uploaded by the PCR detection instrument. The system integrates the experience and knowledge of many agricultural experts in wheat disease and insect pest diagnosis and stores it in the form of a rule base in the server. The judgment and analysis module matches and infers the model output results and detection data with the rules in the expert knowledge base to obtain more accurate and comprehensive disease and insect pest diagnosis conclusions. For example, if the model detects that a certain lesion feature appears on the wheat leaves, and the PCR detection instrument detects the presence of a specific pathogen in the soil, the judgment and analysis module will combine the expert knowledge base on the relationship between the pathogen and the lesion to determine the type of disease and insect pest that the wheat suffers from and its possible severity.

[0066] When the measure generation module gives prevention and control measures for wheat diseases and insect pests based on the results of the judgment and analysis module, it not only provides conventional chemical control methods (such as the recommended types of pesticides, dosages, and usage methods), but also combines biological control and physical control methods. For example, it is recommended to release natural enemies of pests for biological control, or to use physical control methods such as setting up insect-proof nets and insect-attracting lamps. The measure generation module will also make personalized adjustments to the prevention and control measures based on factors such as the growth stage of wheat, local climatic conditions, and soil fertility to ensure the effectiveness and safety of the prevention and control measures. At the same time, the module will generate a detailed implementation plan for prevention and control measures, including prevention and control schedules, operating steps, precautions, etc., to facilitate user implementation.

[0067] The collection unit is used to collect wheat-related pest and disease image data, wheat pest and disease control measures data, and related drugs for controlling wheat pest and disease on the Internet.

[0068] The client is used to display the detection results and specific prevention and control measures, thereby enabling users to manage wheat fields conveniently.

[0069] The client includes a display module for displaying and explaining the data statistics in the resource library. When displaying and explaining the data statistics in the resource library, the display module not only provides a variety of data visualization methods (such as bar charts, line charts, word clouds, heat maps, etc.), but also realizes the interactive function of data visualization. Users can get more detailed information by clicking on the data points in the chart. For example, by clicking on a category in the bar chart of pest and disease image categories, they can view the specific picture samples under that category, the relevant pest and disease descriptions, prevention and control measures and other information. The display module also supports dynamic updating of data. When new data is generated or the data changes on the application server side, the display module of the client can refresh the display content in time, so that users can always grasp the latest wheat field information. In addition, the display module also has a data export function. Users can export data statistics in Excel, PDF and other formats to facilitate further analysis and report generation.

[0070] Second implementation mode: The present invention also discloses an intelligent detection method for wheat pests and diseases, comprising the following steps:

[0071] Step 1: Collection Preparation

[0072] S1: Check and calibrate the drone and PCR testing instrument to ensure that the drone's camera, flight system, soil sampler and wireless transmission module are working properly, and that the PCR testing instrument has sufficient testing reagents, normal testing functions and a stable wireless transmission module;

[0073] S2: Setting the flight parameters of the drone, including flight altitude, speed, shooting interval, sampling area, etc., so as to carry out comprehensive and targeted leaf photo collection and soil sample collection planning for the wheat field;

[0074] Step 2: Data Collection

[0075] S21: The drone flies along a preset route, taking photos of wheat leaves from multiple angles through a camera, and using a soil sampler to collect soil samples at designated locations, which are then automatically sealed in a sample tube.

[0076] S22: During the collection process, the drone uploads the captured leaf photos to the application server in real time through its built-in wireless transmission module

[0077] S23: Detect the type and quantity of pathogens in the collected soil sample through a PCR detection instrument, and upload the data to the application server through a wireless transmission module after the detection is completed;

[0078] Step 3: Data preprocessing and storage

[0079] S31: The leaf photo dataset construction module in the application server receives the leaf photos uploaded by the drone, adjusts the image to a 640*640 square fixed pixel, and normalizes the pixel value of the collected image to between 0 and 1, constructs a rice leaf photo dataset, and stores it in the data storage module

[0080] S32: The data storage module stores the soil sample test data uploaded by the PCR test instrument, and the data obtained by the collection unit from the Internet and uploaded by the user, including wheat-related pest and disease image data, wheat pest and disease control measures data, and relevant drug information for controlling wheat pest and disease, etc.;

[0081] Step 4: Model construction and optimization

[0082] S41: Improve the target detection algorithm model of YOLO based on the lightweight network ShuffleNetv2 and flexible non-maximum suppression Soft-NMS. Replace the feature extraction backbone network Backbone in the YOLO model with the CBRM module in ShuffleNetv2, and replace the non-maximum suppression function NMS in YOLO with Soft-NMS to build a new target detection network model ShuffleNetv2-YOLO.

[0083] S42: Train and optimize the constructed ShuffleNetv2-YOLO model, use the constructed leaf photo dataset for training, and continuously adjust the model parameters during the training process to improve the model's recognition accuracy and recall rate for wheat pest and disease characteristics;

[0084] Step 5: Pest and disease detection and analysis

[0085] S51: Input the leaf photo dataset into the trained target detection network model ShuffleNetv2-YOLO for comparative calculation, and obtain the wheat leaf disease and insect pest detection results output by the model;

[0086] S52: Combine the results of the model output and the data uploaded by the PCR detection instrument, refer to the expert knowledge base and related data information in the data storage module, conduct a comprehensive judgment and analysis on the pathogens of wheat leaf diseases and insect pests, and determine the type, severity and possible spread range of the diseases and insect pests;

[0087] Step 6: Generate and push prevention and control measures

[0088] S61: Based on the results of the judgment and analysis module, generate prevention and control measures for wheat diseases and insect pests, including chemical control (such as pesticide types, dosages, and usage methods), biological control (such as releasing natural enemy insect types and numbers) and physical control (such as setting up insect-proof nets, insect-attracting lights, etc.), and formulate a detailed prevention and control plan

[0089] S62: Push the test results, pest and disease analysis report, and prevention and control measures to the client through the communication module;

[0090] Step 7: Client display and management

[0091] S71: The client receives and displays the test results, pest and disease analysis reports, and specific prevention and control measures pushed by the application server, so that users can intuitively understand the pest and disease situation in the wheat field.

[0092] S72: The display module of the client selects bar graphs, line graphs, word clouds, heat maps, etc. according to the data type and characteristics to display and explain the statistical information such as the number of pictures, the number of news, the number of videos, the categories and number of pest and disease pictures, the type and number of news, the source of news, and the word cloud of news keywords in the database, so as to assist the user to further analyze the relevant information of wheat pests and diseases so that the user can better manage the wheat fields.

[0093] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art, and the model parameters of each electrical appliance are not specifically limited, and conventional equipment can be used.

[0094] 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.

[0095] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent detection system for wheat pests and diseases, characterized in that: include: A collection end, which is used to collect photos of wheat leaves and soil samples from wheat fields; An application server side, which is used to detect the collected soil samples and analyze the leaf photos, and then to determine the pathogens of wheat leaf diseases and insect pests; The client is used to display the detection results and specific prevention and control measures, thereby facilitating users to manage wheat fields.

2. The intelligent detection system for wheat pests and diseases according to claim 1, characterized in that: The collection end includes a drone, a PCR detection instrument and a wireless transmission module integrated in the two respectively. The drone takes multi-angle photos of wheat leaves through its camera and uploads the photos to the application server through the wireless transmission module. The PCR detection instrument is used to detect the types and quantities of pathogens in the collected soil samples, and uploads the data to the application server through the wireless transmission module.

3. The intelligent detection system for wheat pests and diseases according to claim 2, characterized in that: A soil sampler is installed inside the drone, and a sample tube is installed inside the soil sampler. The soil sampler is used to collect soil samples, and the sample tube is used to automatically package the collected soil samples.

4. The intelligent detection system for wheat pests and diseases according to claim 1, characterized in that: The application server includes a data processing unit, a data storage module, a communication module and a collection unit. The data processing unit is used to analyze the soil samples and leaf photos collected by the detection; The data storage module is used to store data obtained from the Internet and uploaded by users; The communication module is used to realize the communication connection between the acquisition end, the client end and the Internet by using the server end; The collection unit is used to collect wheat-related pest and disease image data, wheat pest and disease control measure data, and relevant medicines for controlling wheat pest and disease on the Internet.

5. The intelligent detection system for wheat pests and diseases according to claim 1, characterized in that: The application server also includes a model creation unit, which improves the target detection algorithm model of YOLO based on the lightweight network ShuffleNetv2 and flexible non-maximum suppression Soft-NMS, and constructs a new target detection network model ShuffleNetv2-YOLO.

6. The intelligent detection system for wheat pests and diseases according to claim 5, characterized in that: The specific structure of the target detection network model ShuffleNetv2-YOLO is as follows: the feature extraction backbone network Backbone in the YOLO model is replaced by the CBRM module in the lightweight network ShuffleNetv2, and the non-maximum suppression function NMS in the YOLO is replaced by the flexible non-maximum suppression function Soft-NMS.

7. The intelligent detection system for wheat pests and diseases according to claim 4, characterized in that: The data processing unit includes a blade photo data set construction module, a model calculation comparison module, a judgment analysis module and a measure generation module, and each module is connected through signals; The leaf photo dataset construction module is used to construct a rice leaf photo dataset; The model calculation comparison module is used to input the leaf photo dataset into the target detection network model for comparison calculation; The judgment and analysis module is used to judge and analyze the results output by the model and the data uploaded by the PCR detection instrument; The measure generation module provides prevention and control measures for the diseases and insect pests that appear in wheat based on the analysis results of the judgment and analysis module.

8. The intelligent detection system for wheat pests and diseases according to claim 7, characterized in that: The leaf photo dataset construction module is also used to adjust the image to a square fixed pixel of 640*640, and normalize the pixel value of the collected image to between 0 and 1 to reduce the image inconsistency caused by lighting changes or the influence of other external factors.

9. The intelligent detection system for wheat pests and diseases according to claim 1, characterized in that: The client includes a display module for displaying and explaining the data statistical information in the resource library. The display and explanation of the data statistical information refers to selecting bar charts, line charts, word clouds, and heat maps according to data types and characteristics to display and explain the number of pictures in the database, the number of news, the number of videos, the categories and number of pest and disease pictures, the type and number of news, the source of news, and the news keyword word cloud map.

Images