Chinese rose disease and pest detection method, server, medium and program product

Through the combination of environmental sensors and drones, all-round and all-weather monitoring and intelligent prevention and control of rose pests and diseases has been achieved, solving the problems of incomplete detection and disconnection between prevention and control in the existing technology, and improving detection accuracy and Prevention and treatment effect.

CN118968130APending Publication Date: 2024-11-15SHENZHEN TIMES HORTICULTURE FLORAL CO LTD

Patent Information

Application Number
CN202410974036.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The existing rose pest detection methods are difficult to achieve all-round and all-weather monitoring, especially difficult to detect invisible forms and early pests, and detection and prevention and control are out of touch.

Method used

A variety of information sources are used to detect diseases and pests, and the rose growth parameters and image data are obtained through environmental sensors and drones, and comprehensive judgment is made in combination with image recognition and growth models to achieve all-round and all-weather pest monitoring and intelligent prevention and control.

Benefits of technology

All-round and all-weather monitoring and intelligent prevention and control of rose pests and diseases have been achieved, and various pests and diseases including invisible forms can be detected, and the occurrence and spread of pests and diseases can be effectively suppressed through active prevention and control measures.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118968130A_ABST
    Figure CN118968130A_ABST
Patent Text Reader

Abstract

The invention provides a Chinese rose disease and pest detection method, a server, a medium and a program product, and relates to the field of Internet, the method comprises the following steps: obtaining a plurality of first environment parameter data sent by a plurality of environment sensors, and obtaining image data and second environment parameter data of Chinese rose plants through an unmanned aerial vehicle; determining a potential disease and insect pest area and a disease probabilistic thermodynamic diagram according to the image data and a preset image data set; inputting the first environment parameter data and the second environment parameter data in the potential pest and disease damage area into a pest and disease damage growth model, outputting a judgment result that the current environment is suitable for pest and disease damage growth, and determining a final pest and disease damage comprehensive detection result according to the judgment result, the potential pest and disease damage area and the disease probability thermodynamic diagram. The final comprehensive pest and disease damage detection result at least comprises the name, the occurrence area and the occurrence probability of the pest and disease damage. By implementing the method, plant diseases and insect pests including invisible forms can be comprehensively detected in an all-weather manner.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of Internet, and in particular to a rose disease and insect pest detection method, server, medium and program product. Background Art

[0002] As an important part of my country's agriculture, the prevention and control of flower diseases and pests has always attracted much attention. In recent years, with the development of information technology, online detection technology for flower diseases and pests has emerged, providing new prevention and control methods for the flower industry. Among them, wireless Internet technology is increasingly widely used in agriculture, especially in the detection of flower diseases and pests. Rose is one of the flowers with the most diseases and pests, so its disease and pest detection is of utmost importance.

[0003] The rose pest detection method in the prior art relies on computer vision and image recognition technology. First, sufficient image samples containing various rose pests are collected. Image processing technology is used to pre-process the sample images, remove noise, enhance contrast, etc. The image data set is annotated manually or by machine learning algorithms to obtain the pest category label corresponding to each image, extract the color, texture, shape and other visual features of the image, construct the feature vector of the pest, input the image features and the annotated category label into the supervised learning algorithm, train the classification model, extract the features of the newly input rose pest image, and classify it by the trained model, output the pest category and send it to the management end.

[0004] However, the single image recognition method in traditional methods often sets up image acquisition equipment at a fixed position and only shoots at multiple fixed angles, which makes it difficult to obtain the overall situation of pests and diseases in all directions. On the other hand, some pests and diseases do not have obvious visible forms in the early stage, and it is difficult to achieve early detection of pests and diseases by relying solely on image recognition. In addition, traditional methods rely on visible light images and are difficult to achieve all-weather monitoring. Therefore, it is difficult for traditional methods to comprehensively and all-weather detect pests and diseases, including those in invisible forms. Summary of the invention

[0005] The present application provides a rose disease and insect pest detection method, server, medium and program product for realizing all-round, all-weather monitoring and intelligent detection of rose diseases and insect pests.

[0006] In a first aspect, the present application provides a rose pest and disease detection method, which is applied to a server, and the method includes: obtaining multiple first environmental parameter data sent by multiple environmental sensors, the environmental sensors are arranged on rose plants, and are used to detect the rose growth environment parameters in real time and generate first environmental parameter data; obtaining image data and second environmental parameter data of the rose plant by a drone, the drone is equipped with a camera and a sensor; determining a potential pest and disease area and a disease probability heat map according to the image data and a preset image data set; inputting the first environmental parameter data and the second environmental parameter data in the potential pest and disease area into a pest and disease growth model, and outputting a judgment result that the current environment is suitable for the growth of pests and diseases, the current environment is jointly determined by the first environmental parameter data and the second environmental parameter data, and the pest and disease growth model is constructed based on the environmental data characteristics of multiple historical pest and disease occurrences; after determining the final pest and disease comprehensive detection result based on the judgment result, the potential pest and disease area and the disease probability heat map, it is sent to the receiving end for display, and the final pest and disease comprehensive detection result includes at least the name of the pest and disease, the occurrence area and the probability of occurrence.

[0007] In the above embodiment, the method collects rose growth environment parameters and image data through environmental sensors and drones, uses image recognition, and combines growth model judgment to achieve all-round monitoring and intelligent detection of rose pests and diseases. It avoids the limitations of fixed positions and reliance on a single source of information, and can detect pests and diseases more comprehensively and accurately, including those that are invisible or have just started to hatch. The test results intuitively display the disease area and probability distribution, providing a key basis for subsequent prevention and control, and helping to improve the pertinence and effectiveness of prevention and control.

[0008] In combination with some embodiments of the first aspect, in some embodiments, after the step of inputting the first environmental parameter data and the second environmental parameter data in the potential pest and disease area into the pest and disease growth model and outputting a judgment result that the current environment is suitable for the growth of pests and diseases, it also includes: determining a pest and disease control plan based on the judgment result and a preset pest and disease control plan library; controlling the drone and the preset environmental control equipment to perform environmental adjustment operations according to the pest and disease control plan, so as to keep the environment in the potential pest and disease area in a state that is unfavorable to pests and diseases.

[0009] In the above embodiment, the prevention and control plan can be automatically determined based on the model judgment results and the equipment can be controlled to perform environmental adjustment, so as to effectively inhibit the occurrence and spread of pests and diseases and realize intelligent integrated prevention and control of pests and diseases.

[0010] In combination with some embodiments of the first aspect, in some embodiments, after determining the final comprehensive pest and disease detection result based on the judgment result, the potential pest and disease area and the probabilistic heat map of the disease, and sending it to the receiving end for display, it also includes: planning the drone route according to the final comprehensive pest and disease detection result so that the drone can continue to observe the potential pest and disease area.

[0011] In the above embodiment, the drone can be made to continuously observe key areas according to the detection results to obtain more comprehensive image data and environmental parameters, which is conducive to improving the accuracy of detection of pests and diseases.

[0012] In combination with some embodiments of the first aspect, in some embodiments, after the step of inputting the first environmental parameter data and the second environmental parameter data in the potential pest and disease area into the pest and disease growth model and outputting the judgment result that the current environment is suitable for the growth of pests and diseases, it also includes: if the judgment result is that the current environment is suitable for the growth of pests and diseases, a pest and disease warning prompt is sent to the receiving end, and corresponding prevention and control measures are determined according to the preset pest and disease control program library and sent to the receiving end.

[0013] In the above embodiment, the server can send early warning prompts to the user in a timely manner and recommend corresponding prevention and control measures to achieve intelligent disease and insect pest control.

[0014] In combination with some embodiments of the first aspect, in some embodiments, after the final comprehensive pest and disease detection result is determined based on the judgment result, the potential pest and disease area and the probabilistic heat map of the disease, and the step of sending it to the receiving end for display, it also includes: controlling the receiving end to highlight the potential pest and disease area; after receiving the detailed information request instruction, sending detailed information of the pest and disease to the receiving end, the detailed information includes at least a description of the pest and disease, an assessment of the degree of damage and an analysis of the causes of the pest and disease.

[0015] In the above embodiments, the detection results can be displayed intuitively, and detailed pest information can be provided according to user needs, so that users can understand the pest situation conveniently.

[0016] In combination with some embodiments of the first aspect, in some embodiments, after the step of acquiring image data and second environmental parameter data of rose plants through a drone, it also includes: sending the first environmental parameter data, the second environmental parameter data and the image data to an expert end; receiving diagnostic information sent by the expert end, and sending the diagnostic information to a receiving end for display.

[0017] In the above embodiment, the server will send the collected image data and environmental parameter data to the expert end for comprehensive analysis. The expert end can be composed of senior plant pathologists, who can diagnose rose diseases and insect pests by combining images, environmental parameters and other information, determine the name of the disease and insect pests, the degree of harm and other information, and send the diagnosis results back to the server, which then transmits the expert diagnosis information to the user receiving end for display. In this way, the knowledge of professionals can be used for more accurate diagnosis, which helps to improve the effect of disease and insect pest detection.

[0018] In combination with some embodiments of the first aspect, in some embodiments, the step of determining the potential pest and disease area and the disease probability heat map based on the image data and a preset image data set specifically includes: training a deep learning model based on the image data set, the image data set refers to images containing various types of rose pests and diseases and having pixel-level annotations in the pest and disease area; inputting the image data into the deep learning model, judging whether each image pixel in the image data is classified as a pest and disease, and obtaining a classification judgment result; determining the probability size of pests and diseases in multiple segmented images based on the classification judgment result; determining the potential pest and disease area based on the pest and disease probability size; determining the disease probability heat map based on the pest and disease probability size.

[0019] In the above embodiment, the server uses a deep learning model to perform image segmentation and classification, which can improve the accuracy of disease area identification and avoid missed detection and false detection. Compared with the traditional rule-based image processing algorithm, the deep learning model can learn complex feature representations through a large number of labeled image training, and has better disease identification effects for different rose varieties and different environments. Pixel-level segmentation can locate the disease range more precisely, and the output disease probability heat map intuitively reflects the distribution of disease severity. This can comprehensively improve the quality of disease and insect pest detection and positioning, and provide an accurate basis for subsequent diagnosis and prevention.

[0020] In a second aspect, an embodiment of the present application provides a server, comprising: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the server to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions, which, when executed on a server, enables the server to execute the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, comprising instructions, which, when executed on a server, enable the server to execute the method described in the first aspect and any possible implementation of the first aspect.

[0023] It is understandable that the server provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the method provided in the embodiment of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Since a variety of information sources are used for pest and disease detection, rose growth parameters and image information are obtained through environmental sensors, drone images, growth models and other means, so all-round monitoring of rose pests and diseases can be achieved, avoiding the limitation of relying solely on a single information source and a fixed location, and effectively solving the problem that existing technologies are difficult to fully cover pest and disease detection, thereby realizing all-weather, all-round detection of various pests and diseases.

[0025] 2. Due to the judgment method combining image recognition and pest and disease growth model, various pests and diseases including invisible forms can be detected, even those that are difficult to observe in the early stages can be judged. This effectively solves the problem that the existing technology relies on visible light images and is difficult to detect invisible pests and diseases, thereby realizing accurate detection of various pests and diseases including invisible forms.

[0026] 3. Since the method of actively controlling pests and diseases based on the judgment results is adopted, active intervention in the pest and disease areas can be achieved. By controlling the drones and environmental equipment to adjust the environmental parameters, the problem of disconnection between detection and prevention in the existing technology is effectively solved, thereby realizing the close integration of pest and disease detection and prevention and intelligent prevention and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a flow chart of a method for detecting rose pests and diseases in an embodiment of the present application; Figure 2 This is another flow chart of the rose pest detection method in the embodiment of the present application; Figure 3 It is a schematic diagram of a physical device structure of a server in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification of the present application, the singular expressions "one", "a kind of", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more of the listed items.

[0029] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0030] For ease of understanding, an application scenario of the relevant method is introduced below.

[0031] Master Wang is responsible for the maintenance of roses in a park. Various roses are planted in the park, and some diseases and pests often occur. In order to detect and prevent rose diseases and pests in time, Master Wang set up multiple image monitoring devices in the park. These devices are fixed in designated locations and take photos of roses at regular intervals every day. The obtained pictures will be automatically uploaded to the system background. The system uses the image recognition model to determine the type of diseases and pests in the picture and send the test results to Master Wang's user end. A few days ago, Master Wang found that a large area of ​​roses in the northwest corner of the park showed symptoms suspected of gray mold. However, that area happened to be in the blind spot of the monitoring equipment. There had never been any problems in the past, and the system did not detect any abnormalities. When Master Wang discovered it, the disease was already quite serious. If it had been discovered earlier, the loss would have been much smaller. Master Wang also found that the damage of some cabbage moth larvae was difficult to see from the surface of the leaves in the initial stage. The serious infestation caused by them when they grew up would trigger the system's detection alarm, but it was too late. Another time, Master Wang found that the detection effect of the system would be significantly worse on rainy days or at night. One day, it was raining heavily, and the system did not send out any pest alarms. After the rain, when Master Wang checked the flowerbed, he found that a large area was severely eaten by butterfly larvae, causing heavy losses. It turned out that it had been raining for the past two days, and the monitoring camera could not capture clear images, so the butterfly larvae hidden behind the leaves could not be identified.

[0032] It can be seen that there are limitations in detecting rose diseases and pests by relying solely on fixed image monitoring equipment and simple recognition models, and it is difficult to achieve comprehensive, accurate and intelligent detection results.

[0033] By adopting the rose pest detection method in the embodiment of the present application, by acquiring environmental sensor data and drone image data, combined with image recognition and growth model judgment, all-round, all-weather monitoring and intelligent detection of rose pests and diseases are achieved, which not only avoids the limitation of relying on a single information source and a fixed position, but also can detect various pests and diseases including invisible forms. The following introduces the scene using the rose pest detection method in the present application.

[0034] Following the above example, Master Wang first installed dozens of small environmental monitoring devices in the flower beds in key areas of the park. The equipment contains temperature and humidity sensors, light sensors, etc., which can monitor the environmental indicators of pest incubation in real time. In addition, an eight-rotor drone is set up, which can cruise over the rose park and take high-definition rose images through an onboard infrared camera. The drone is controlled by a program to ensure that each flower bed is photographed in all directions. The obtained environmental data and images will be transmitted to the background server. The server uses deep learning algorithms to segment and classify the images, find out the potential areas of lesions in the images, and determine whether it is currently in a period prone to pests and diseases based on environmental parameters. The server can comprehensively judge the possible areas and severity of pests and diseases, generate a thermal distribution map of the disease, and send the test results to the user terminal after determining the specific pests and diseases.

[0035] After using the system, Master Wang found that the detection effect of pests and diseases has been significantly improved. Once the system determined the peak period of bacterial wilt based on temperature and humidity, and found the potential affected area in the image, Master Wang immediately found the prevention and control plan recommended by the historical model, and promptly controlled the drone and sprayer for prevention and control, thus avoiding the large-scale spread of bacterial wilt. With continuous monitoring and early warning, Master Wang has achieved all-weather, no-dead-angle intelligent detection of rose pests and diseases, which not only improves the detection quality, but also provides key support for subsequent prevention and control work, greatly reducing physical fatigue.

[0036] For ease of understanding, the following describes the process of the method provided by this implementation in combination with the above scenario. Figure 1 , is a flow chart of the rose disease and insect pest detection method in the embodiment of the present application.

[0037] S101, obtaining a plurality of first environmental parameter data sent by a plurality of environmental sensors, where the environmental sensors are arranged on rose plants and are used to detect the growth environmental parameters of the roses in real time and generate the first environmental parameter data; In this step, the server will receive the rose growth environment parameter data uploaded from multiple environmental sensors through the network. The environmental sensor can be a temperature and humidity sensor, a light sensor, or other sensors that can detect the environment, which is not limited here. The sensor is pre-arranged on the setting position of the rose plant, monitors the environmental condition information such as the temperature, humidity, and light of the rose growth in real time, and transmits the detected environmental parameters to the server in real time through the wireless network module in the form of digital signals. The server will continue to receive these environmental parameter data that are continuously transmitted, and the environmental parameter data sent to the server by the sensor here is the first environmental parameter data.

[0038] For example, a temperature and humidity sensor detects the current temperature and humidity every 10 minutes, and sends the collected temperature and humidity values ​​to the server via WiFi. The server will receive temperature and humidity data from the sensor every 10 minutes. At the same time, the server will also obtain various parameter information from other temperature and humidity sensors and light sensors at the same frequency. The server will distinguish the data sent by different sensors based on the sensor's identification information. All environmental data will be summarized and recorded to provide environmental reference for subsequent disease judgment.

[0039] By using multi-point and multi-type environmental sensors, we can monitor the rose growth environment in more dimensions and obtain more comprehensive and detailed environmental parameter information. Real-time transmission of environmental data can also reflect the dynamic changes of the environment, providing data support for the next step of disease identification and growth environment judgment.

[0040] S102, acquiring image data of rose plants and second environmental parameter data by using a drone, where the drone is equipped with a camera and a sensor; In this step, the server will control the drone to collect all-round images of the rose plants according to the set program requirements. The drone is equipped with a digital camera, which can be installed under the drone or in other locations that are convenient for collecting images according to user needs. This is not limited here, so that high-definition rose images can be taken; at the same time, the drone also has built-in temperature and humidity sensors and other sensors that can detect the environment, which can detect environmental parameters during flight. The environmental parameter data sent by the drone to the server is the second environmental parameter data.

[0041] The server will control the drone to fly automatically according to the preset cruise route, so that it will spiral up to cover the entire rose monitoring area, hover stably at a specific height, and use the camera to take all-round images of the rose. The drone can take pictures of roses from multiple different angles to obtain multi-angle rose image data with a three-dimensional sense. At the same time, the server will instruct the drone to start the onboard environmental sensor when it moves along the cruise route, and collect the second set of environmental parameter information in real time. This set of parameters reflects the air environment and can supplement the first set of environmental sensor data deployed near the ground.

[0042] The image data and environmental data collected by the drone are also transmitted to the server in real time, providing the information source required for image recognition and environmental judgment. The drone's cruise frequency can be executed regularly or started when receiving a server command to achieve all-weather monitoring.

[0043] S103, determining a potential pest and disease area and a disease probability heat map according to the image data and a preset image data set; In this step, the server will analyze the high-definition rose images uploaded by the drone based on deep learning technology, identify potential pests and diseases in the images, and determine the disease probability of each area, and finally output the disease area division map and thermal probability map. Specifically, the server pre-collects image data sets containing various rose pests and diseases, and uses manual annotation or machine learning algorithms to achieve pixel-level segmentation of the image, marking the specific range of each type of disease. Then, these annotated images are used to train the deep neural network model to learn feature extraction and image classification methods. For the newly input drone rose images, the server will be divided into multiple local areas, and then input into the pre-trained deep learning model in turn, and output whether there is a disease in each area, as well as the category and probability of the disease. Finally, the server will integrate the detection results of all local areas to generate a disease area division map and thermal probability map for the entire image.

[0044] This can effectively identify the location distribution and severity of diseases in the image, realize intelligent monitoring of the health status of roses, and provide guidance for subsequent prevention and control work.

[0045] S104, inputting the first environmental parameter data and the second environmental parameter data in the potential pest area into a pest growth model, and outputting a judgment result that the current environment is suitable for pest growth, wherein the current environment is jointly determined by the first environmental parameter data and the second environmental parameter data, and the pest growth model is constructed based on the environmental data characteristics of various historical pest occurrences; In this step, the server will use the first environmental parameter data and the second environmental parameter data obtained previously as input and feed them into the pre-established pest and disease growth model. The model will comprehensively consider the current environmental parameters to determine whether the environment is suitable for the generation and spread of pests and diseases.

[0046] Specifically, the server will collect a large amount of environmental data on the occurrence of rose pests and diseases in different regions and at different times. These data can come from research conducted by scientific research institutions or from farmers' records. The server will use machine learning algorithms to learn the environmental characteristics of different pests and diseases from a large amount of historical data and train a pest and disease growth model. The model combines multiple environmental factors such as temperature, humidity, and light to determine the pest and disease risk coefficient under certain environmental conditions. For the currently acquired environmental parameters, the server will input the first environmental parameter data and the second environmental parameter data into the model together. The model will calculate a comprehensive environmental score to determine whether the environment is prone to induce pests and diseases or promote the rapid growth and spread of pests and diseases. Under normal environmental conditions, the model outputs a low score; when environmental changes make pests and diseases easy to occur, the model will output a higher score. The server will set a threshold suitable for the growth of pests and diseases based on the score. When the score exceeds the threshold, it is determined that the current environment is suitable for the growth of pests and diseases. In this way, the pest and disease growth model can accurately determine the impact of the environment on pests and diseases, and provide a basis for subsequent prevention and control measures. The model can also be continuously optimized and updated through continuous learning.

[0047] In some embodiments, when the server determines that the current environment is very suitable for the generation and spread of a certain pest based on the first environmental parameter, the second environmental parameter and the pest growth model, the server needs to immediately send a pest warning prompt to the receiving end to remind the user to pay attention to the risk of the pest. Specifically, the server will query the name, occurrence probability and degree of harm of the pest from the preset database, and generate a pest warning prompt, including the title, occurrence probability, impact range and other contents of the disease. For example, "Attention, according to monitoring, lotus white silk disease has appeared in the vegetable garden area, with a probability of 90%, which may spread rapidly and seriously harm crops." Then, the server will immediately push the warning information to the receiving end, such as the administrator's mobile phone or computer, in the form of a pop-up window or audio prompt to attract the administrator's attention. At the same time, the server will also query the preset pest control knowledge base according to the pest and select the corresponding prevention and control measures. The knowledge base is pre-entered by pathology experts, including various disease prevention and control methods, drug selection and usage methods. The server will refer to the knowledge base to determine the most appropriate prevention and control measures in the current environment.

[0048] For example, for lotus white wilt disease, the server will find out that its prevention and control measures are "in the early stage, you can spray 20 times diluted cypermethrin wettable powder for prevention and control", so the server will push the prevention and control measures to the receiving end and inform the user to take the above measures to prevent and control pests and diseases.

[0049] In this way, through the push of early warning prompts and prevention and control suggestions, users can be helped to understand the situation in a timely manner, take prevention and control measures quickly, and curb the occurrence and spread of pests and diseases.

[0050] S105. After the final comprehensive pest and disease detection result is determined based on the judgment result, the potential pest and disease area and the probabilistic heat map of the disease, it is sent to the receiving end for display. The final comprehensive pest and disease detection result at least includes the name of the pest and disease, the occurrence area and the occurrence probability.

[0051] In this step, the server will comprehensively utilize the previously obtained judgment results, heat maps and regional information to determine the final pest and disease detection results, and send the results to the receiving end for display on the interface.

[0052] Specifically, the server uses a pre-trained deep learning model to perform semantic segmentation and classification on rose images to obtain the disease category prediction results and probabilities for each area. For example, the probability that area A is bacterial wilt is 0.8. The server inputs the pre-built pest and disease growth model based on the environmental parameter data to obtain the risk score of the current environment for various pests and diseases. For example, the risk score of the current environment for bacterial wilt is 0.75. The server fuses the image recognition results and the model judgment results. If the judgment results of the two are consistent, the category is determined as the final result. For example, if the probability and risk score of bacterial wilt are both high values ​​in the above example, the detection result is determined to be bacterial wilt. The image recognition results and the growth model judgment results can also be fused according to the set ratio to determine the most likely name of the pest and disease. Similarly, pests and diseases are determined for other areas in this way, and the occurrence area of ​​pests and diseases is determined in combination with the heat map, and the occurrence probability of different sub-areas in the area is calculated. Finally, the server will generate the final comprehensive pest and disease detection results and send them to the receiving end. The results will clearly state the names of the detected pests and diseases, and mark the range of pest and disease occurrence in the graphical interface. At the same time, different color heat maps are used in the area to represent different occurrence probabilities. The receiving end can be the manager's computer or mobile terminal such as a mobile phone. After obtaining the test results sent by the server, the receiving end will visually highlight the pest and disease area in the interface and provide text descriptions, so that managers can clearly and intuitively understand the severity and distribution of the current rose pests and diseases, providing a basis for subsequent prevention and control work.

[0053] In some embodiments, the server can highlight all potential pest and disease areas detected in the final comprehensive pest and disease detection results, and use eye-catching colors to frame the pest and disease area range. For example, a red rectangle is used to frame the bacterial wilt area, and a blue circle is used to circle the gathering area of ​​tobacco stink bugs. At the same time, the server will control the interface of the receiving end to automatically zoom in and out the detection result picture to highlight the highlighted disease area and make it the focus of the display interface. Secondly, the server provides a detailed information query function. The receiving end user can select the disease area and send a query request to the server. According to the request, the server will retrieve relevant information about the disease from the database, including: description of the clinical manifestations of the disease, such as the onset cycle, symptom changes, etc.; medical evaluation, such as the degree of harm is divided into light, medium and heavy, and the transmission speed is divided into slow, medium and fast; expert analysis of the cause of the disease, such as related insect activities, climate humidity, etc. The server can organize the above three aspects of information into a detailed text or audio file and send it to the receiving end. It is also possible to select only key information to push according to the processing capacity of the receiving end. Detailed information can help users fully understand the disease situation and improve the prevention and control effect.

[0054] In summary, the server can realize intelligent and refined pest and disease detection through multi-source data analysis, and the detection results can be clearly communicated to end users to improve identification efficiency.

[0055] In some embodiments, an Internet of Things platform can be developed, which connects various sensors and devices to achieve remote monitoring and management of the rose growth environment, and the server builds and maintains the Internet of Things platform. Specifically, the server will pre-set and connect a variety of different types of environmental sensors, including temperature and humidity sensors, light sensors, soil moisture sensors, etc. These sensors are arranged around the rose plants to monitor parameters such as temperature, humidity, light, and soil in real time. The server will also connect sprinkler irrigation equipment, ventilation equipment, shading equipment, etc. to achieve automatic control of the rose growth environment. All devices are connected to the server via a wireless network, and the server centrally manages the device status. The platform database stores real-time data and operating status of various devices. The server provides a management interface, and users can remotely view environmental parameter values, equipment operating status, historical data statistics, and other information, and can also issue control instructions, such as adjusting the sprinkler irrigation volume or opening the shading device. The platform realizes fully automated monitoring and intelligent management of the rose growth environment. Users can monitor the growth status of roses, obtain real-time data, manage planting plans, etc. through the application. Specifically, the server can also develop an APP application for mobile phones and tablets. After users download the APP, they can see the real-life picture or model of the rose, and it is equipped with real-time data from various sensors, such as the current temperature of 28°C and humidity of 65%. Users can understand the environmental status at any time. Users can also switch to view historical data curves for different periods to evaluate the impact of environmental changes on the growth of roses. The mobile APP also has a planting calendar function. Users can make and modify planting plans such as watering, fertilizing, and pruning for roses, and the application will automatically remind users to complete the operation according to the plan. Some equipment operations can also be triggered remotely through the APP, such as turning on the water pump for irrigation. The mobile APP realizes convenient planting management, and users can keep track of the growth dynamics of roses at any time. In short, the Internet of Things platform and mobile APP developed by the server realize all-round monitoring and intelligent management of Xiangyang roses, which can remotely monitor environmental parameters, obtain growth status in time, make planting plans and automate operations, providing users with very convenient planting management services, which helps to improve the growth quality of roses.

[0056] In some embodiments, after determining the potential pest and disease area, the drone can be asked to focus on detecting this area. Specifically, the server can obtain information such as the name, severity and precise area of ​​the pest and disease from the test results. Taking bacterial wilt as an example, the test results show that there is bacterial wilt in the tulip garden in the southeast area of ​​the park. The probability of bacterial wilt in multiple segmented images is greater than 15%. It is judged that the area is highly prevalent and needs to be monitored. After obtaining the result, the server will automatically find the coordinate information of the flower garden area and plan the cruise route of the drone. The server will give priority to setting the drone route to pass through the potential area of ​​bacterial wilt, and can instruct the drone to hover over the area for an additional period of time, such as circling 5 times more, to ensure that the roses in the area are taken in all directions. In this way, clearer and more comprehensive image data can be obtained, which is conducive to continuous monitoring of the development of diseases in the area. At the same time, the server can also adjust the cruising altitude and speed of the drone in the area according to the severity of the disease. For severe areas, the flight altitude is reduced and the hovering time is increased to obtain higher resolution images. These images are richer in content and are conducive to model detection and feature extraction. For areas with less severe diseases, normal cruise parameters can be maintained. In addition, the server can also determine areas with poor image quality based on the detection results, which may be severely blocked or the disease is not obvious. For these areas, it is also necessary to lower the altitude and speed to obtain clearer images. This can continuously improve the monitoring quality of key areas.

[0057] By dynamically adjusting the drone route, we can achieve continuous observation of key disease areas and actively optimize data collection to monitor the development and changes of diseases more timely and accurately, providing support for subsequent prevention and control work. If the test results are updated in real time, the server can also dynamically adjust the drone route to ensure rapid response to newly discovered disease areas.

[0058] After combining the above embodiments, the method provided by this embodiment is further described in more detail below. Figure 2 , is another flow chart of the rose disease and insect pest detection method in the embodiment of the present application.

[0059] S201, training a deep learning model according to the image data set, where the image data set refers to images containing various types of rose pests and diseases and having pixel-level annotations in the pest and disease areas; First, the server needs to collect images containing various common rose diseases and pests as a data set. These images can be obtained by taking photos by horticultural experts and collecting them by web crawlers. Professionals can use image segmentation tools to divide each image in the data set at the pixel level, manually mark the exact range of each disease, and obtain accurate semantic segmentation results. For example, for images of tobacco stink bugs, the annotator will accurately divide all pixel areas where the tobacco stink bugs are located. These segmented images are used as training data sets for the server. Then, the server will use these annotated image data sets to train deep learning models such as convolutional neural networks. The deep learning model can automatically learn the visual features of different diseases in the image, such as color, shape, texture, etc., through rotation training. After sufficient iterative training, the model has acquired the ability to identify various rose diseases. These pre-trained deep learning models are saved by the server and are ready for subsequent image recognition predictions.

[0060] S202, inputting the image data into the deep learning model, determining whether each image pixel in the image data is classified as a pest, and obtaining a classification determination result; During the monitoring process, the server will control the drone to continuously collect images of the rose plants. After obtaining the rose pictures, the server will use these high-definition images as input and feed them into the pre-trained deep learning model one by one. The deep learning model will output the prediction results for each image, that is, what category each pixel belongs to, and whether the pixel is a normal part or a disease. For example, for a 100*100 pixel image, the model will determine that pixel 1 is not diseased, pixel 2 is bacterial wilt, and so on, until the classification results of all 100,000 pixels in the image are output. The server will store these prediction results and record the locations of the pixels judged to be diseased in each image, in preparation for the next step of area determination.

[0061] S203, determining the probability of pests and diseases in the multiple segmented images according to the classification judgment result; In this step, the server will count the number of pixels judged as diseased in each image, calculate the proportion of different types of diseased pixels to the total number of pixels in the image, and obtain the probability of each type of disease in the image. Specifically, for an image of 500500 pixels, the judgment result of the deep learning model shows that 120 pixels in the image are judged as bacterial wilt and 300 pixels are judged as gray mold. The server will calculate the probability of bacterial wilt disease pixels accounting for the total number of pixels in the image as 120 / 500500=0.048, or 4.8%; the probability of gray mold disease pixels is 300 / 500500=0.12, or 12%. The server will count the probability of each category of disease for each image in turn and save it in the database. It should be noted that each original image will be divided into multiple small blocks by the server at equal intervals to form multiple segmented images, so that more detailed results can be obtained. For example, an original image of 4000*4000 pixels will be equally divided into 16 segmented images of 1000*1000 pixels. The server will perform the above processing on each segmented image to obtain detailed disease probability statistics for each segmented image. Afterwards, the server can summarize the results of different segmented images as needed to obtain the overall probability of each category of disease in the original image. For example, among the 16 segmented images of an original image, 10 segmented images have a higher probability of bacterial wilt, then it can be judged that the original image as a whole has a more serious bacterial wilt. The receiving end user can also choose to view the specific statistics of the original image or a segmented image area. These disease probability results can intuitively reflect the severity of the disease.

[0062] S204, determining the potential pest area according to the pest probability; In this step, the server will judge and determine whether the segmented image belongs to a potential pest and disease area according to the set threshold and the size of the disease probability in each segmented image. Specifically, the server will set a threshold for each type of disease. For example, the judgment threshold for bacterial wilt is set to 5% of the total pixels of the image due to disease, and the judgment threshold for gray mold is set to 8% of the total pixels of the image due to disease. When the probability of bacterial wilt in a segmented image exceeds 5%, the image is judged to belong to a potential area for bacterial wilt; when the probability of gray mold exceeds 8%, it is judged to be a potential area for gray mold. The same applies to other types of diseases. It should be noted that the setting of the threshold should take into account factors such as the resistance of plants to different diseases and the degree of disease damage. Proper threshold setting can avoid missed reports and false alarms.

[0063] The server can also determine multiple closely related adjacent segmented images as a whole potential area. For example, if five adjacent segmented images all detect a high probability of bacterial wilt, it can be determined that they together constitute a potential spread area of ​​bacterial wilt. This can better locate the area where the disease is concentrated. Finally, the server can obtain accurate potential area information for all types of diseases in the original image.

[0064] S205. Determine a probability heat map of the disease according to the probability of the disease and insect pest.

[0065] In this step, the server will color each segmented image according to its disease probability and generate a heat map reflecting the distribution of the disease. Specifically, for a certain disease, such as bacterial wilt, the server will preset a disease probability-color mapping table, and use different colors for visualization according to different bacterial wilt probability values, such as 0-5% for green, 5-10% for yellow, 10-15% for orange, 15-20% for red, and greater than 20% for black. In this way, a segmented image is mapped to the corresponding color according to its statistical probability of bacterial wilt. For example, an image with a probability of 7% is colored yellow, and an image with a probability of 13% is colored orange. The server will color all the segmented images of the original image one by one, and finally form a heat map of multi-color regional distribution. Through color rendering, the severity and spread of bacterial wilt in the image are intuitively reflected.

[0066] Similarly, the server will generate a corresponding probability heat map for each disease. These heat maps are superimposed to form a panoramic view of the disease distribution marked with different colors. The receiving end user can choose to display the heat map of a certain type of disease, or display the summary heat map of all categories, so that the areas with the most serious diseases and pests can be clearly seen at a glance. These probability heat maps provide an important reference for subsequent precise prevention and control.

[0067] In the embodiments of the present application, due to the use of deep learning-based image recognition technology, combined with precise pixel-level annotation and a rich training sample data set, and considering the probability of occurrence of different diseases as the basis for judgment, it is possible to accurately judge the disease area and severity in the rose image, thereby achieving refined detection of the distribution of rose disease areas and severity.

[0068] In some embodiments, when the server determines that a certain pest has appeared in the rose through image recognition and model judgment, or when the environmental parameter analysis believes that a certain pest may occur, the server will start the active prevention mode, and control the environment by controlling the drone and environmental equipment to adjust the environment to control the environmental conditions for pest control. Specifically, when it is determined that pests and diseases have occurred or there is a trend of occurrence, the server will first query the relevant prevention and control schemes in the preset pest control knowledge base according to the type of pests and diseases. The knowledge base collects various prevention and control measures for pests and diseases, including environmental adjustment, drug spraying and other methods. For example, for bacterial wilt, the knowledge base will recommend reducing regional humidity and temperature to slow down the growth of diseases. The server will select the best environmental adjustment scheme under the current circumstances, and then calculate the optimal environmental target values ​​of environmental parameters according to the scheme, such as humidity control at 40-50% and temperature control at 20-25°C. Then, the server will control the execution of the scheme. It will instruct drones to carry specific functional payloads and spray agricultural chemicals, such as fungicides or nitrogen fertilizers, over the pest-infested areas to quickly change the microenvironment of the diseased areas and inhibit the growth of pathogens. The server will also send control instructions to environmental control equipment (such as sprinkler systems, thermal devices, etc.) to adjust their operating modes to achieve the best temperature and humidity environment in the diseased areas, such as reducing humidity and increasing ventilation, to further optimize the growth environment and inhibit the rebound of diseases.

[0069] At the same time, the server will continue to track the real-time data of the environmental sensors to determine whether the environmental parameters meet the optimal environmental target values. If there is a deviation, the server will dynamically fine-tune the control parameters of the drone and environmental equipment until the environmental data meets the requirements of the prevention and control environment, thereby maintaining a growth environment that is very unfavorable to the growth of diseases in a long-term and stable manner, curbing the generation and spread of diseases. In addition, the server can also trigger the drone's environmental processing tasks in advance based on this. When the monitoring parameters indicate a trend of occurrence but the disease has not yet appeared on a large scale, it will actively carry out environmental processing to further reduce the risk of disease outbreaks.

[0070] Through the coordinated operation of intelligent server-controlled drones and environmental equipment, environmental parameters in pest-affected areas can be adjusted quickly and specifically, pests and diseases can be proactively controlled, the occurrence and spread of diseases can be effectively suppressed, and losses to rose growth can be reduced.

[0071] In some embodiments, blockchain technology can also be used to track the entire process of roses from planting to sales to ensure transparency and traceability. The server builds and maintains a blockchain tracking system. Specifically, when the rose plant emerges, the server can generate a blockchain account as the digital identity ID of the plant, and record the initial information such as the source of the seed, the breeder, and the location and write it into the blockchain account book. During the growth of the rose, all kinds of data collected by environmental sensors and drones are also summarized under the ID to form a complete rose growth file, such as temperature and humidity curves, image records, etc., to ensure that the entire growth process is traceable. When it comes to the harvesting stage, the server will generate a harvesting block and write data such as yield, harvest date, and person in charge. In the production and sales stage, the person in charge of each circulation node of the rose product needs to confirm with a digital signature. The server will generate a new block to store transaction and transportation details, including product description, quantity, quality inspection report, carrier, recipient and other information. When the rose product finally reaches the consumer end, the consumer can scan the code to read the product blockchain and clearly see the complete process of the rose from planting to sales. In this way, using blockchain to track the entire process of roses can ensure the reliability of the source of rose products and the transparency of their circulation, thereby enhancing consumers' sense of trust.

[0072] Blockchain technology can also be used to protect the intellectual property rights of new varieties and prevent infringement. The server builds an intellectual property protection system based on blockchain. Specifically, after the cultivation of a new rose variety is completed, the server will generate a block on the blockchain account book, in which the key information of the new variety is recorded, including the name of the new variety, the breeder, the main characteristics, etc. At the same time, there will be a timestamp on the block, recording the exact time when the new variety information is written into the blockchain. This timestamp can serve as proof of the digital copyright of the new variety.

[0073] After the new variety is officially put into production and sales, the server will continue to monitor whether there is any plagiarism in the market. Specifically, the server will collect data from third-party channels, such as content posted by users in flower forums and social media. If it is found that someone else claims to have cultivated a variety that is very similar to the new variety, it can be determined that there is potential plagiarism. Once signs of plagiarism are identified, the server can start the preset copyright protection mechanism accordingly. The server can query the blockchain ledger and prove the priority of the new variety through the timestamp on the new variety block, indicating that the breeder of the new variety owns the intellectual property rights of the new variety. Then, the server can send a warning message to the plagiarist in the form of a smart contract, requiring it to stop the infringement. At the same time, the server can also send a prompt to the management end to let the administrator pay attention to possible plagiarism in the market so that further protection measures can be taken. For customers who purchase the new variety of rose yarn, when the transaction is completed, the server will automatically generate a digital copyright transaction certificate on the blockchain and issue it to the customer. In this way, the roses that the customer subsequently grows and propagates the new variety also have a traceable source and copyright protection to prevent the product from being impersonated by the plagiarist.

[0074] By providing reliable copyright protection through blockchain technology, we can fully safeguard the rights and interests of new variety breeders and effectively prevent others from infringing and misappropriating new varieties, thereby encouraging the cultivation of more new varieties and promoting the progress and prosperity of rose varieties.

[0075] The following describes the server in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , is a schematic diagram of a physical device structure of a server in an embodiment of the present application.

[0076] It should be noted that Figure 3 The structure of the server shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0077] like Figure 3 As shown, the server includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage part 308 to the random access memory (RAM) 303, such as executing the method described in the above embodiment. In RAM 303, various programs and data required for system operation are also stored. CPU 301, ROM 302 and RAM 303 are connected to each other through bus 304. Input / output (I / O) interface 305 is also connected to bus 304.

[0078] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, a button switch, etc.; an output section 307 including a liquid crystal display (LCD) and an audio output device, an indicator light, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed so that a computer program read therefrom is installed into the storage section 308 as needed.

[0079] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 309, and / or installed from a removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, various functions defined in the present invention are performed.

[0080] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0081] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings.

[0082] Specifically, the server of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the rose disease and insect pest detection method provided in the above embodiment is implemented.

[0083] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the server described in the above embodiment; or may exist independently without being assembled into the server. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of the server, the server implements the rose pest detection method provided in the above embodiment.

[0084] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0085] As used in the above embodiments, the term "when..." may be interpreted to mean "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted to mean "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0086] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. A rose pest detection method, applied to a server, characterized in that: The method comprises: Acquire a plurality of first environmental parameter data sent by a plurality of environmental sensors, wherein the environmental sensors are arranged on the rose plants and are used to detect the rose growth environmental parameters in real time and generate the first environmental parameter data; Acquiring image data of rose plants and second environmental parameter data by using a drone, wherein the drone is equipped with a camera and a sensor; Determine potential pest and disease areas and a disease probability heat map based on the image data and a preset image data set; Inputting the first environmental parameter data and the second environmental parameter data in the potential pest area into a pest growth model, and outputting a judgment result that the current environment is suitable for pest growth, wherein the current environment is jointly determined by the first environmental parameter data and the second environmental parameter data, and the pest growth model is constructed based on the environmental data characteristics of various historical pest occurrences; After the final comprehensive pest and disease detection result is determined based on the judgment result, the potential pest and disease area and the probabilistic heat map of the disease, it is sent to the receiving end for display. The final comprehensive pest and disease detection result at least includes the name of the pest and disease, the occurrence area and the probability of occurrence.

2. The method according to claim 1, characterized in that After the step of inputting the first environmental parameter data and the second environmental parameter data in the potential pest area into the pest growth model and outputting the judgment result that the current environment is suitable for the growth of pests, the method further includes: Determine a pest control plan based on the judgment result and a preset pest control plan library; According to the pest control plan, the drone and the preset environmental control equipment are controlled to perform environmental adjustment operations so that the environment in the potential pest area is maintained in a state that is unfavorable to pests.

3. The method according to claim 1, characterized in that After determining the final comprehensive pest and disease detection result according to the judgment result, the potential pest and disease area and the pest and disease probability heat map, and sending it to the receiving end for display, the method further includes: The drone route is planned according to the final comprehensive pest and disease detection result so that the drone can continuously observe the potential pest and disease area.

4. The method according to claim 1, characterized in that: After the step of inputting the first environmental parameter data and the second environmental parameter data in the potential pest area into the pest growth model and outputting the judgment result that the current environment is suitable for the growth of pests, the method further includes: If the judgment result is that the current environment is suitable for the growth of pests and diseases, a pest and disease warning prompt is sent to the receiving end, and corresponding prevention and control measures are determined according to a preset pest and disease control program library and sent to the receiving end.

5. The method according to claim 1, characterized in that After determining the final comprehensive pest and disease detection result according to the judgment result, the potential pest and disease area and the pest and disease probability heat map, and sending it to the receiving end for display, the method further includes: The control receiving end highlights the potential pest and disease area; After receiving the detailed information request instruction, the detailed information of the pests and diseases is sent to the receiving end, and the detailed information at least includes the description of the pests and diseases, the evaluation of the degree of damage and the analysis of the causes of the pests and diseases.

6. The method according to claim 1, characterized in that After the step of obtaining the image data of the rose plant and the second environmental parameter data by the drone, the method further includes: Sending the first environmental parameter data, the second environmental parameter data and the image data to an expert terminal; The diagnostic information sent by the expert end is received, and the diagnostic information is sent to the receiving end for display.

7. The method according to claim 1, characterized in that The step of determining the potential pest and disease area and the disease probability heat map according to the image data and the preset image data set specifically includes: Training a deep learning model based on the image data set, wherein the image data set refers to images containing various types of rose pests and diseases and having pixel-level annotations in the pest and disease areas; Inputting the image data into the deep learning model, determining whether each image pixel in the image data is classified as a pest, and obtaining a classification determination result; Determine the probability of pests and diseases in multiple segmented images according to the classification judgment results; Determine the potential pest and disease area according to the pest and disease probability; The disease probability heat map is determined according to the probability of the disease and insect pests.

8. A server, characterized in that: The server includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the server to execute the method described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a server, the server is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a server, the server is caused to execute the method according to any one of claims 1 to 7.

Citation Information

Patent Citations

  • Unmanned aerial vehicle on-board real-time image processing method and system for crops

    CN109977924A

  • Intelligent monitoring method and system for forest diseases and insect pests

    CN113822225A

  • Crop disease and insect pest detection method adapted to rainy environment

    CN114842300A

  • Rosa chinensis seedling growing information and environment monitoring method

    CN115496300A

  • Crop disease and pest detection method and system based on multi-modal deep learning

    CN118258769A

Cited By

  • Integrated system for fertilization and pest control for emblica cultivation

    CN119744624A

  • Rice disease and pest monitoring method and system

    CN120147863A

  • Method and system for intelligently monitoring and accurately preventing and treating plant diseases and insect pests in traditional Chinese medicinal material standard base

    CN120806423A