Crop detection method, device, equipment, storage medium and program product

Abstract

The invention discloses a crop detection method and device, equipment, a storage medium and a program product, and the method belongs to the technical field of image processing and crop management. The method comprises the following steps: acquiring a first image representing a whole area of a target farmland corresponding to field patrol time through a satellite remote sensing technology, and then acquiring a second image representing a to-be-detected crop by using an unmanned aerial vehicle technology based on the first image, the to-be-detected crop being a crop in the target farmland; and finally, based on the second image, the field patrol time and a pre-constructed farmland crop database, the field patrol efficiency can be improved while the crop growth risk can be identified in time.

Description

Technical Field

[0001] This application belongs to the field of image processing and crop management technology, and in particular relates to a crop detection method, apparatus, equipment, storage medium and program product. Background Technology

[0002] In agricultural production, effective crop management is a core element in ensuring crop yields and reducing growth risks. Among these, field inspections are a crucial step in crop management, with the core objective of collecting data on crop growth status through on-site observation and using this data to assess the actual growth condition of the crops.

[0003] Currently, the main method relies on manual field inspections, where managers observe crop growth in real time by walking around the fields, manually collect relevant data, and conduct analysis to determine the crop growth status.

[0004] However, when farmland is cultivated on a large scale, manual field inspections require a significant investment of manpower and time, resulting in low overall efficiency. Furthermore, the quality of data collection is significantly affected by subjective factors such as the experience and judgment of management personnel, and in large-scale inspection scenarios, oversights and missing records are prone to occur, making it difficult to guarantee the accuracy and completeness of the collected data, and consequently, failing to identify crop growth risks in a timely manner. Summary of the Invention

[0005] This application provides a crop detection method, apparatus, equipment, storage medium, and program product that can improve field inspection efficiency while timely identifying crop growth risks.

[0006] In a first aspect, embodiments of this application provide a method for detecting crops, the method comprising:

[0007] Obtain patrol mission instructions, which include: patrol time and target farmland location;

[0008] Based on the location of the target farmland and the patrol time, the first image representing the overall area of ​​the target farmland corresponding to the patrol time is obtained by satellite remote sensing technology.

[0009] Based on the first image, a second image representing the crop to be detected is obtained using UAV technology. The crop to be detected is the crop in the target farmland.

[0010] Based on the second image, the field inspection time, and the pre-built farmland and crop database, a growth risk analysis is performed on the crops to be tested, generating a growth risk assessment result for the crops to be tested. The growth risk includes: lateral growth risk and longitudinal growth risk. The farmland and crop database includes: farmland location information, crop information associated with the farmland location information, and historical reference images of each growth stage corresponding to the crop information.

[0011] In some possible implementations, based on the second image, field inspection time, and a pre-built farmland crop database, a growth risk analysis is performed on the crop to be tested, generating a growth risk assessment result for the crop to be tested, including:

[0012] Based on the location of the target farmland and the information of the crop to be detected, reference images are retrieved from the farmland and crop database. The reference images include: horizontal reference images and vertical reference images. Horizontal reference images refer to historical reference images that are identical to the information of the crop to be detected and whose growth stage matches the field inspection time. Vertical reference images refer to reference images that are located in the same target farmland location as the crop to be detected and whose collection time is within a preset time window before the field inspection time.

[0013] Identify the image feature points of the second image, the horizontal reference image, and the vertical reference image, and extract the feature vectors corresponding to the image feature points;

[0014] Based on the similarity algorithm, the first similarity between the feature vector of the second image and the feature vector of the horizontal reference image, and the second similarity between the feature vector of the second image and the feature vector of the vertical reference image are calculated.

[0015] The first similarity is compared with the horizontal similarity threshold, and the second similarity is compared with the vertical similarity threshold to generate the growth risk assessment result.

[0016] In some possible implementations, the method further includes, before comparing the first similarity with the lateral similarity threshold:

[0017] Extract the first historical health image dataset from the farmland and crop database that has the same information as the crop to be detected and whose growth stage matches the field inspection time;

[0018] Calculate the horizontal similarity values ​​between images in the first historical health image dataset, and perform statistical distribution analysis on the horizontal similarity values;

[0019] The horizontal similarity threshold is calculated based on the statistical distribution of the horizontal similarity values.

[0020] In some possible implementations, the method includes comparing the second similarity with the longitudinal similarity threshold before:

[0021] Based on each first historical health image in the first historical health image dataset, extract the second historical health image corresponding to the first historical health image from the farmland and crop database, whose collection time is within a preset time window before it;

[0022] Calculate the longitudinal similarity value between each first historical health image and its corresponding second historical health image, and perform statistical distribution analysis on the longitudinal similarity values;

[0023] The vertical similarity threshold is calculated based on the statistical distribution of the vertical similarity values.

[0024] In some possible implementations, based on the location of the target farmland and the patrol time, a first image representing the overall area of ​​the target farmland corresponding to the patrol time is acquired using satellite remote sensing technology, including:

[0025] Based on the location of the target farmland and the patrol time, satellite images corresponding to the patrol time are acquired, and the target farmland image is cropped from the satellite images based on the location of the target farmland.

[0026] An edge detection algorithm is used to segment the target farmland image, extract the edge information of the target farmland, and generate a first image with georeferenced information based on the edge information of the target farmland.

[0027] In some possible implementations, based on the first image, a second image representing the crop to be detected is acquired using drone technology, including:

[0028] Based on the geographic reference information recorded in the first image, the drone flight path is planned, and the drone is controlled to take pictures along the flight path to obtain the image sequence of the target farmland.

[0029] Image stitching is performed on the image sequence of the target farmland to obtain UAV imagery of the target farmland;

[0030] Image analysis algorithms are used to process UAV images, identify and segment crop areas, and generate a second image.

[0031] In some possible implementations, before performing growth risk analysis on the crop to be tested based on the second image, field inspection time, and a pre-built farmland crop database, and generating a growth risk assessment result for the crop to be tested, the following steps are also included:

[0032] Obtain information on the locations of multiple farmlands and crops;

[0033] The location of farmland and crop information are associated and mapped to generate structured farmland attribute information;

[0034] Farmland attribute information is uploaded to a cloud server and saved to a farmland and crop database.

[0035] In some possible implementations, after performing growth risk analysis on the crop to be tested based on the second image, field inspection time, and a pre-built farmland crop database, and generating a growth risk assessment result for the crop to be tested, the following steps are also included:

[0036] The location of the target farmland, the first image, the second image, the patrol time, and the growth risk assessment results are correlated and mapped to form a sample of farmland and crop data.

[0037] Farmland and crop data samples are uploaded to a cloud server, and based on farmland attribute information, the farmland and crop data samples are associated and stored in a farmland and crop database.

[0038] Secondly, embodiments of this application provide a crop detection device, the device comprising:

[0039] The first acquisition module is used to acquire field patrol task instructions, which include: field patrol time and target farmland location.

[0040] The second acquisition module is used to acquire a first image representing the overall area of ​​the target farmland corresponding to the patrol time using satellite remote sensing technology, based on the location of the target farmland and the patrol time.

[0041] The third acquisition module is used to acquire a second image representing the growth area of ​​the crop to be detected based on the first image using UAV technology. The crop to be detected is the crop in the target farmland.

[0042] The analysis module is used to perform growth risk analysis on the crops to be tested based on the second image, the field inspection time, and a pre-built farmland and crop database, and generate growth risk assessment results for the crops to be tested. The growth risks include: lateral growth risk and longitudinal growth risk. The farmland and crop database includes: farmland location information, crop information associated with the farmland location information, and historical reference images of each growth stage corresponding to the crop information.

[0043] Thirdly, embodiments of this application provide an electronic device, the device comprising:

[0044] A processor and a memory storing computer program instructions; a crop detection method that implements any one of the above when the processor executes the computer program instructions.

[0045] Fourthly, embodiments of this application provide a computer storage medium storing computer program instructions, which, when executed by a processor, implement any of the above-mentioned crop detection methods.

[0046] Fifthly, embodiments of this application provide a computer program product in which instructions, when executed by the processor of an electronic device, enable the electronic device to perform any of the above-mentioned crop detection methods.

[0047] This application discloses a crop detection method, apparatus, device, storage medium, and program product. The method involves acquiring a field patrol task instruction, then, based on the target farmland location and patrol time in the instruction, using satellite remote sensing technology to acquire a first image representing the overall area of ​​the target farmland corresponding to the patrol time. Based on the first image, a second image representing the crop to be detected is acquired using unmanned aerial vehicle (UAV) technology. The crop to be detected is the crop within the target farmland. Finally, based on the second image, the patrol time, and a pre-constructed farmland and crop database, a growth risk analysis is performed on the crop to be detected, generating a growth risk assessment result. Compared to existing technologies, manual field patrols are time-consuming and inefficient, and the accuracy and completeness of the collected data are difficult to guarantee, making it impossible to identify crop growth risks in a timely manner. This application addresses this issue by receiving a field patrol task instruction and automatically acquiring a first image of the entire farmland area based on the target farmland location and patrol time included in the instruction, using satellite remote sensing technology. Then, based on the geographical information in the first image, a drone is controlled to automatically collect a second image representing the crop growth area. By integrating satellite remote sensing and drone technologies, field patrols can be completed efficiently, and a large number of farmland and crop images can be collected, reducing the time cost of manual field patrols and avoiding the problems of inaccurate and incomplete data collection during manual field patrols. Therefore, after acquiring the second image, feature comparison and similarity analysis are performed with health history images in the farmland and crop database to achieve accurate assessment of crop growth risks, enabling timely identification of crop growth risks and reducing crop losses. Attached Figure Description

[0048] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 A schematic flowchart of a crop detection method according to an embodiment of this application is shown;

[0050] Figure 2 A flowchart illustrating a crop detection method according to another embodiment of this application is shown;

[0051] Figure 3 A flowchart illustrating a crop detection method according to another embodiment of this application is shown;

[0052] Figure 4 A flowchart illustrating a crop detection method according to another embodiment of this application is shown;

[0053] Figure 5This paper presents a schematic diagram of the overall process of a crop detection method according to this application;

[0054] Figure 6 A schematic diagram of the structure of the crop detection device provided in the embodiment of this application is shown;

[0055] Figure 7 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0056] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0057] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0058] It should be noted that the acquisition, storage, use, and processing of data in this application embodiment all comply with the relevant provisions of national laws and regulations.

[0059] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0060] Currently, field inspections are primarily conducted manually. Managers observe the crop's production status in real time through on-site inspections, manually collecting and analyzing relevant data. However, when dealing with large-scale farmland, manual inspections require significant manpower and time, resulting in low efficiency. Furthermore, in large-area inspection scenarios, oversights and missing records are prone to occur, making it difficult to guarantee the accuracy and completeness of the collected data. This, in turn, affects the timely identification and early warning of crop growth risks.

[0061] To address the problems of existing technologies, this application provides a method, apparatus, equipment, storage medium, and program product for crop detection. By receiving a field patrol task instruction, and based on the target farmland location and patrol time included in the instruction, a first image of the entire farmland area is automatically acquired using satellite remote sensing technology. Then, based on the geographical information in the first image, a drone is controlled to automatically collect a second image representing the crop growth area. By integrating satellite remote sensing and drone technologies, efficient, automated, and large-scale farmland image acquisition is achieved, reducing the time cost of manual field patrols. Finally, after acquiring the second image, feature comparison and similarity analysis are performed between it and health history images in a farmland and crop database to accurately assess crop growth risks. Based on the risk assessment results, corresponding control measures can be provided in a timely manner, reducing crop losses.

[0062] The following is a description of a crop detection method provided in the embodiments of this application.

[0063] Figure 1 A schematic flowchart of a crop detection method according to an embodiment of this application is shown. Figure 1 As shown, the method may include the following steps:

[0064] S101: Obtain patrol task instructions, which include: patrol time and target farmland location.

[0065] In this embodiment of the application, the field patrol task instruction includes: patrol time and target farmland location. The patrol time refers to the specific time point at which the patrol task is executed. In one example, the target farmland location can be obtained through satellite positioning technology and represented in geographic coordinates. For example, the geometric boundary of the target farmland can be delineated using a series of ordered, closed vertex latitude and longitude coordinate sequences to represent the target farmland location.

[0066] In another example, patrol task instructions can be issued in several ways: Patrol task instructions can be manually triggered by managers, for example, by actively selecting the target farmland location and setting the patrol time through a mobile terminal or management platform interface, thereby instantly generating and issuing the patrol task instruction; patrol task instructions can be automatically triggered based on a preset cycle, for example, managers can configure a fixed patrol plan for each farmland in the system, and the system will automatically generate a patrol task instruction when the predetermined time arrives, realizing routine unattended patrol tasks; patrol task instructions can be triggered based on an event mechanism, for example, when the system receives meteorological early warning information and a severe weather forecast, it will automatically generate an emergency patrol task instruction to obtain the latest farmland images in time before the disaster occurs, providing a comparative basis for subsequent disaster impact assessment and damage analysis.

[0067] S102: Based on the location of the target farmland and the patrol time, acquire the first image representing the overall area of ​​the target farmland corresponding to the patrol time using satellite remote sensing technology.

[0068] In this embodiment, satellite images corresponding to the field patrol time are acquired based on the patrol time. Then, the location of the target farmland is obtained through satellite positioning technology and spatially registered and cropped with satellite images obtained through satellite remote sensing technology, thereby obtaining a first image that completely matches the field patrol task instructions in both time and space dimensions.

[0069] In one example, to provide a clear and accurate spatial basis for subsequent UAV field patrol path planning and crop analysis, a specific implementation of step S102 is as follows:

[0070] Based on the location of the target farmland and the patrol time, satellite images corresponding to the patrol time are acquired, and the target farmland image is cropped from the satellite images based on the location of the target farmland.

[0071] In this embodiment, satellite images corresponding to the patrol time are acquired from multiple satellite images according to the patrol mission instruction. Further, the satellite images are cropped based on the target farmland location to obtain the image region of the target farmland location, i.e., the target farmland image. In one example, the target farmland image is obtained by cropping the satellite images using vector data composed of a series of ordered closed vertex latitude and longitude coordinates.

[0072] An edge detection algorithm is used to segment the target farmland image, extract the edge information of the target farmland, and generate a first image with georeferenced information based on the edge information of the target farmland.

[0073] In this embodiment, due to the low resolution of satellite imagery, the cropped target farmland image typically still includes non-target areas such as field ridges, roads, and surrounding vegetation. Therefore, by further employing an edge detection algorithm to perform image segmentation processing on the target farmland image, edge information of the farmland area is identified and extracted. Based on the extracted edge information, image extraction is performed on the target farmland image to generate a first image containing only the target farmland area and retaining geographic reference information. In one example, the edge detection algorithm can be the Sobel operator, the Canny operator, or a deep learning-based edge detection model.

[0074] In one example, after obtaining the first image, it can be visually verified and adjusted online via a mobile terminal or management platform interface. By adjusting the edge information of the target farmland, the obtained first image becomes more accurate. Simultaneously, the corrected first image can be stored in a training database for periodic iterative optimization of the edge detection algorithm, thereby continuously improving the accuracy of automatic boundary delineation.

[0075] In this embodiment, satellite images are cropped based on the target farmland location and patrol time to obtain the target farmland image corresponding to the target farmland location. Then, image segmentation processing is performed on the target farmland image based on an edge detection algorithm to effectively eliminate non-target farmland areas. The generated first image provides a clear and accurate spatial basis for the subsequent UAV patrol path. Furthermore, the entire process requires no manual intervention. By receiving patrol task instructions, satellite images can be automatically retrieved and processed. Through spatial registration and edge detection algorithms, the division and extraction of the target farmland area can be quickly completed, avoiding the heavy workload of traditional manual on-site surveying and improving operational efficiency and quality.

[0076] S103: Based on the first image, a second image representing the crop to be detected is obtained using UAV technology. The crop to be detected is the crop in the target farmland.

[0077] In this embodiment of the application, based on the obtained first image, a high-definition image of the target farmland is captured using drone technology to obtain a second image of the crops to be detected growing in the target farmland.

[0078] In one example, to obtain a high-resolution image of the crop to be detected, step S103 can be implemented as follows:

[0079] Based on the geographic reference information recorded in the first image, the drone flight path is planned, and the drone is controlled to take pictures along the flight path to obtain an image sequence of the target farmland.

[0080] In this embodiment of the application, based on the geographic reference information recorded in the first image, in one example, the geographic reference information recorded in the first image refers to the vector data composed of a series of ordered closed vertex latitude and longitude coordinates corresponding to the first image. Based on the vector data corresponding to the first image, the flight path of the UAV is planned, and the UAV is controlled to take pictures along the flight path to obtain the image sequence of the target farmland.

[0081] In one example, a multispectral imaging device can be mounted on a drone to automatically capture images of the target farmland as the drone flies along its flight path.

[0082] Image stitching is performed on the image sequence of the target farmland to obtain UAV imagery of the target farmland.

[0083] In this embodiment, image stitching is performed on the image sequence of the target farmland to obtain UAV imagery of the target farmland. In one example, based on feature point extraction and matching algorithms, an image matrix of adjacent target farmlands can be calculated, and all images can be aligned and fused to generate UAV imagery of the target farmland.

[0084] In another example, edge computing can be used during drone flight to stitch together images of the target farmland in real time using the drone's onboard computer, thereby improving the efficiency of obtaining drone images.

[0085] Image analysis algorithms are used to process UAV images, identify and segment crop areas, and generate a second image.

[0086] In this embodiment of the application, an image analysis algorithm can be used to process the drone image to identify and segment the crop area. In one example, the image analysis algorithm can be an edge detection algorithm. The edge detection algorithm is used to process the drone image to identify and segment the crop area in the farmland.

[0087] In this embodiment, the flight path of the UAV is automatically planned using the geographic reference information recorded in the first image. The UAV is then controlled to automatically collect image sequences of the target farmland along the flight path, and the image sequences of the target farmland are stitched together to form a UAV image. Finally, image processing is performed on the UAV image to identify and segment the crop areas, generating a second image. By replacing traditional manual field inspections with UAV technology, large-area crop image collection tasks can be completed efficiently and automatically. Furthermore, the crop images acquired by UAVs have high resolution and strong coverage, significantly improving the efficiency and quality of crop image acquisition and effectively avoiding subjective errors and manual data entry mistakes introduced during manual collection.

[0088] S104: Based on the second image, field inspection time, and a pre-built farmland and crop database, perform growth risk analysis on the crop to be tested and generate growth risk assessment results for the crop to be tested. Growth risks include: lateral growth risk and longitudinal growth risk.

[0089] In this embodiment of the application, the farmland and crop database records farmland location information, crop information associated with the farmland location information, and historical reference images of each growth stage corresponding to the crop information. Based on the field inspection time, historical reference images for lateral growth analysis and historical reference images for longitudinal growth analysis can be extracted from the farmland and crop database. Finally, the currently collected second image is compared and calculated with the extracted reference images to achieve accurate assessment of crop growth risk and generate corresponding growth risk assessment results.

[0090] Among them, horizontal growth analysis refers to comparing the current growth status of crops with the growth status of the same period in history to determine whether the crops deviate from the normal range in the current growth cycle; vertical growth analysis refers to continuously comparing the growth status of the same crop at different time points to determine whether the crops have experienced any abnormalities during the growth process.

[0091] In one example, to further improve the planning and execution efficiency of field patrol tasks, the following steps are included before step S104:

[0092] Obtain multiple farmland locations and crop information.

[0093] In this embodiment of the application, information on multiple farmland locations and crops is first collected. In one example, the crop information includes: crop type, expected growth cycle, sowing date, etc.

[0094] By associating farmland location with crop information, structured farmland attribute information is generated.

[0095] In this embodiment, farmland location and crop information are mapped together to generate a unique identifier for each farmland unit, thereby constructing structured farmland attribute information. In one example, the structured information can be stored and managed using a relational database to support efficient relational queries.

[0096] Farmland attribute information is uploaded to a cloud server and saved to a farmland and crop database.

[0097] In this embodiment of the application, farmland attribute information is uploaded to a cloud server and stored in a farmland and crop database, providing a structured query basis for subsequent field patrol tasks initiated based on farmland location or crop information.

[0098] In this embodiment of the application, before conducting field inspections, farmland location and crop information are collected in advance, and the farmland location and crop information are associated and mapped to generate structured farmland attribute information. Then, the farmland attribute information is stored in the farmland and crop database, which can provide an accurate data index foundation for subsequent automated field inspection tasks, and at the same time establish a reliable data query foundation for crop growth risk assessment.

[0099] In one example, to construct a sustainable and optimized agricultural detection method, after step S104, the following steps are also included:

[0100] The location of the target farmland, the first image, the second image, the time of the field inspection, and the results of the growth risk assessment are correlated and mapped to form a sample of farmland and crop data.

[0101] In this embodiment of the application, the location of the target farmland is used as a unique identifier to associate and map the target farmland location, the first image, the second image, the field inspection time, and the growth risk assessment results to form a farmland crop data sample.

[0102] Farmland and crop data samples are uploaded to a cloud server, and based on farmland attribute information, the farmland and crop data samples are associated and stored in a farmland and crop database.

[0103] In this embodiment of the application, farmland and crop data samples are uploaded to a cloud server. Based on the farmland location identifier contained in the farmland attribute information, the association and storage of farmland and crop data samples with the corresponding farmland units in the farmland and crop database are automatically completed to update the farmland and crop database.

[0104] In one example, the association results between farmland crop data samples and farmland units can be visually verified and manually corrected through the user interface of a mobile terminal or management platform. All confirmed or corrected association data will be saved to a dedicated training database for periodic iterative optimization and model training of the automatic association algorithm, thereby continuously improving the association accuracy.

[0105] In this embodiment, after each field inspection, the target farmland location, first image, second image, inspection time, and growth risk assessment results are correlated and mapped to form a farmland crop data sample. This sample is then uploaded to a cloud server and, based on farmland attribute information, stored in a farmland crop database to update the database. This enables dynamic updating and expansion of the database. By continuously accumulating farmland crop data samples, not only is complete traceability and reusable management of historical farmland data achieved, but also dynamic optimization and adaptive adjustment of horizontal and vertical similarity thresholds can be implemented, thereby continuously improving the accuracy of crop risk identification.

[0106] In this embodiment, a field patrol task instruction is obtained, and then, based on the target farmland location and patrol time in the field patrol task instruction, a first image representing the overall area of ​​the target farmland corresponding to the patrol time is obtained using satellite remote sensing technology. Based on the first image, a second image representing the crop to be detected is obtained using UAV technology. The crop to be detected is the crop in the target farmland. Finally, based on the second image, the patrol time, and a pre-built farmland and crop database, a growth risk analysis is performed on the crop to be detected, and a growth risk assessment result of the crop to be detected is generated. Compared to existing technologies, manual field patrols are time-consuming and inefficient, and the accuracy and completeness of the collected data are difficult to guarantee, making it impossible to identify crop growth risks in a timely manner. This application addresses this issue by receiving a field patrol task instruction and automatically acquiring a first image of the entire farmland area based on the target farmland location and patrol time included in the instruction, using satellite remote sensing technology. Then, based on the geographical information in the first image, a drone is controlled to automatically collect a second image representing the crop growth area. By integrating satellite remote sensing and drone technologies, field patrols can be completed efficiently, and a large number of farmland and crop images can be collected, reducing the time cost of manual field patrols and avoiding the problems of inaccurate and incomplete data collection during manual field patrols. Therefore, after acquiring the second image, feature comparison and similarity analysis are performed with health history images in the farmland and crop database to achieve accurate assessment of crop growth risks, enabling timely identification of crop growth risks and reducing crop losses.

[0107] Figure 2 A schematic flowchart of a crop detection method according to another embodiment of this application is shown. Figure 2 As shown above, in the above Figure 1 Based on the illustrated embodiment, one specific implementation of step S104 is as follows:

[0108] S201: Based on the location of the target farmland and the information of the crops to be detected, retrieve reference images from the farmland and crop database.

[0109] In this embodiment, reference images are retrieved from a farmland and crop database based on the target farmland location and the information of the crop to be detected. These reference images include horizontal reference images and vertical reference images. Horizontal reference images are historical reference images that match the information of the crop to be detected and whose growth stage matches the field inspection time. Vertical reference images are reference images located in the same target farmland as the crop to be detected and whose acquisition time falls within a preset time window before the field inspection time.

[0110] In one example, matching growth stage with field inspection time means determining the current growth stage of the crop to be inspected based on the field inspection time, and then querying the field crop database for a horizontal image with the same growth stage as the current stage; or directly obtaining a historical image from the same period as the current field inspection time as a horizontal reference image.

[0111] In one example, the longitudinal reference image can be a historical reference image acquired within a preset time window prior to the field inspection time. The length of the preset time window can be flexibly set according to the actual monitoring accuracy requirements. Alternatively, the longitudinal reference image can be a historical reference image at the same crop growth stage corresponding to the field inspection time. Since crop growth is a continuous dynamic process, only images within the preset time window prior to the field inspection time can be compared to effectively capture the recent growth trend of crops. If the time interval is too long, the crops may have entered different growth stages, and their morphology and physiological state may have changed significantly, failing to accurately reflect the continuous dynamic process of the current stage, thus affecting the accuracy of the change analysis.

[0112] S202: Determine the image feature points of the second image, the horizontal reference image, and the vertical reference image, and extract the feature vectors corresponding to the image feature points.

[0113] In this embodiment, image feature points of the second image, the horizontal reference image, and the vertical reference image are determined, and feature vectors corresponding to the image feature points are extracted. In one example, feature vectors corresponding to the second image, the horizontal reference image, and the vertical reference image can be extracted based on Scale-Invariant Feature Transform (SIFT).

[0114] S203: Based on the similarity algorithm, calculate the first similarity between the feature vector of the second image and the feature vector of the horizontal reference image, and the second similarity between the feature vector of the second image and the feature vector of the vertical reference image.

[0115] In this embodiment of the application, a first similarity between the feature vector of the second image and the feature vector of the horizontal reference image is calculated based on a similarity algorithm, and a second similarity between the feature vector of the second image and the feature vector of the vertical reference image is calculated based on a similarity algorithm.

[0116] In one example, the similarity algorithm is the cosine similarity algorithm, which is used for two vectors. The similarity between two vectors can be represented by calculating the cosine of their included angle. The formula is as follows:

[0117]

[0118] in, These represent the feature vectors to be compared. This represents the magnitude of the eigenvector A; This represents the magnitude of the eigenvector B; This represents the dot product of vectors A and B; This represents the cosine similarity between vector A and vector B.

[0119] When the angle between vectors A and B is 0, the two vectors are in the same direction, indicating that vectors A and B have the highest similarity, and the cosine similarity is 1.

[0120] When the angle between vectors A and B is 90°, the two vectors are perpendicular and their cosine similarity is 0.

[0121] When the angle between vectors A and B is 180°, the two vectors are in opposite directions, and the cosine similarity is -1.

[0122] S204: Compare the first similarity with the horizontal similarity threshold, and compare the second similarity with the vertical similarity threshold to generate a growth risk assessment result.

[0123] In this embodiment of the application, the first similarity is compared with the horizontal similarity threshold to perform a horizontal growth risk assessment and generate a horizontal risk assessment result; and the second phase velocity is compared with the vertical similarity threshold to perform a vertical growth risk assessment and generate a vertical risk assessment result.

[0124] In one example, if the first similarity is not within the horizontal similarity threshold, a horizontal risk is determined, and a risk assessment result containing the horizontal growth anomaly type is generated; if the second similarity is not within the vertical similarity threshold, a vertical risk is determined, and a risk assessment result containing the vertical growth anomaly type is generated; if both horizontal and vertical risks are triggered simultaneously, a risk fusion judgment is performed, a comprehensive risk assessment result is generated, and the risk level is increased.

[0125] For example, when the first similarity is less than or equal to the horizontal similarity threshold, it may indicate horizontal risks such as pests or diseases or nutritional deficiencies; when the second similarity is less than or equal to the vertical similarity threshold, it may indicate vertical risks such as slow growth; when the second similarity is greater than or equal to the vertical similarity threshold, it may indicate vertical risks such as excessively rapid growth.

[0126] In this embodiment, based on the location of the target farmland and the information of the crop to be detected, a horizontal reference image and a vertical reference image are retrieved from the farmland and crop database. Then, based on the feature vector between the second image and the horizontal reference image, a first similarity is calculated using a similarity algorithm. Based on the feature vector between the second image and the vertical reference image, a second similarity is calculated using a similarity algorithm. The first similarity is compared with a horizontal similarity threshold to generate a horizontal growth risk, and the second similarity is compared with a vertical similarity threshold to generate a vertical growth risk. By simultaneously introducing two comparison mechanisms, horizontal and vertical, the deviation between the current growth state and the historical health state can be judged from a spatial dimension, and abnormal changes in growth rate can be captured from a temporal dimension. This effectively avoids misjudgment or omission caused by single-dimensional analysis and significantly improves the comprehensiveness and accuracy of risk identification.

[0127] Figure 3 A schematic flowchart of a crop detection method according to another embodiment of this application is shown. Figure 3 As shown above, in the above Figure 2 Based on the illustrated embodiment, before step S204, the following steps are also included:

[0128] S301: Extract the first historical health image dataset from the farmland and crop database. The data must be identical to the information of the crop to be detected and the growth stage must match the field inspection time.

[0129] In this embodiment, multiple historical health images are extracted from a farmland crop database. These images contain information identical to the crop to be detected and match the growth stage with the field inspection time. In one example, a historical health image refers to a historical image in which both the lateral and longitudinal growth risk assessment results are healthy.

[0130] S302: Calculate the horizontal similarity value between each image in the first historical health image dataset, and perform statistical distribution analysis on the horizontal similarity value.

[0131] In this embodiment of the application, the horizontal similarity value between each image in the first historical health image dataset is calculated. In one example, the horizontal similarity value can also be calculated using the cosine similarity algorithm, and a statistical distribution analysis is performed on the horizontal similarity value. In one example, the statistical distribution analysis includes calculating the mean and standard deviation of multiple horizontal similarity values.

[0132] S303: Calculate the horizontal similarity threshold based on the statistical distribution of horizontal similarity values.

[0133] In this embodiment of the application, the horizontal similarity threshold is calculated based on the statistical distribution of the horizontal similarity values. In one example, the lower limit of the horizontal similarity threshold is the mean minus the standard deviation, and the upper limit of the horizontal similarity threshold is the mean plus the standard deviation.

[0134] In this embodiment of the application, a first historical health image dataset with the same information as the crop to be detected and whose growth stage matches the field inspection time is extracted from the farmland crop database. The horizontal similarity value between each image in the first historical health image dataset is calculated, and the horizontal similarity value is statistically distributed. Based on the statistical distribution analysis, the horizontal similarity threshold is dynamically determined, which avoids the limitation of relying on manual experience to set a fixed threshold. This allows the threshold to truly reflect the normal performance range of crops at a specific growth stage.

[0135] Figure 4 A schematic flowchart of a crop detection method according to another embodiment of this application is shown. Figure 3 As shown above, in the above Figure 3 Based on the illustrated embodiment, before step S204, the following steps are also included:

[0136] S401: Based on each first historical health image in the first historical health image dataset, extract the second historical health image corresponding to the first historical health image from the farmland and crop database, whose acquisition time is within a preset time window before it.

[0137] In this embodiment of the application, based on each first historical health image in the first historical health image dataset, another historical health image that was collected earlier than the image and is located within a preset time window before it is extracted from the farmland and crop database to form a second historical health image, thereby obtaining multiple pairs of historical health images. Each pair of historical health images includes: a first historical health image and its corresponding second historical health image.

[0138] S402: Calculate the longitudinal similarity value between each first historical health image and its corresponding second historical health image, and perform statistical distribution analysis on the longitudinal similarity value.

[0139] In this embodiment of the application, the longitudinal similarity value between each first historical health image and its corresponding second historical health image is calculated. In one example, the longitudinal similarity value can also be calculated using the cosine similarity algorithm. Statistical distribution analysis is performed on the obtained multiple longitudinal similarities. In one example, the statistical distribution analysis includes calculating the mean and standard deviation of the multiple longitudinal similarities.

[0140] S403: Calculate the vertical similarity threshold based on the statistical distribution of vertical similarity values.

[0141] In this embodiment of the application, the vertical similarity threshold is obtained based on the statistical distribution of the vertical similarity values. In one example, the lower limit of the vertical similarity threshold is the mean of the vertical similarity values ​​minus the standard deviation, and the upper limit of the vertical similarity threshold is the mean of the vertical similarity values ​​plus the standard deviation.

[0142] In this embodiment, based on each first historical healthy image in the first historical healthy image dataset, a second historical healthy image corresponding to the first historical healthy image and collected within a preset time window prior to it is extracted from the farmland and crop database. Then, the longitudinal similarity value between each first historical healthy image and its corresponding second historical healthy image is calculated, and the longitudinal similarity value is statistically distributed. Finally, based on the statistical distribution of the longitudinal similarity value, a longitudinal similarity threshold is calculated. By analyzing the similarity change pattern of historical healthy images within the preset time window, a longitudinal similarity threshold that conforms to the normal growth rate of crops is calculated, avoiding judgment bias caused by relying on fixed thresholds or human experience, and significantly improving the accuracy of abnormal detection in the growth process.

[0143] Figure 5 This diagram illustrates the overall process of a crop detection method according to this application. S501: Automated field patrol and crop information collection. Farmland and crop information are collected using image processing technology. By utilizing satellite remote sensing and drone technology to replace traditional manual field patrols, field patrols can be completed efficiently, and a large number of farmland and crop images can be collected, significantly improving patrol efficiency. S502: Crop growth monitoring and risk warning. Crop growth risks are identified based on similarity calculations. Similarity calculations are performed on crop images obtained through field patrol instructions, horizontal reference images matching the current crop information and growth stage in the farmland and crop database, and vertical reference images close to the current time. The calculated first similarity is compared with a horizontal similarity threshold to determine horizontal growth risk, and the calculated second similarity is compared with a vertical similarity threshold to determine vertical growth risk. Finally, based on the horizontal and vertical growth risks, abnormal crop production conditions are identified, and timely control measures are provided to reduce crop losses. S503: Data storage and database construction. Data results are accumulated, and historical sample data is managed. By linking and integrating multi-source data generated during each field inspection, a structured farmland and crop database sample is formed. This not only supports historical queries and tracing, but also serves as an important training sample for model evolution. Through continuous accumulation and learning, it drives the dynamic optimization and adaptive adjustment of horizontal and vertical similarity thresholds, thereby continuously improving the accuracy of crop risk identification.

[0144] Based on the crop detection method provided in the above embodiments, this application also provides a specific implementation of a crop detection device. Please refer to the following embodiments.

[0145] First see Figure 6 , Figure 6 This illustration shows a schematic diagram of the structure of a crop detection device provided in an embodiment of this application. The crop detection device 600 provided in this embodiment includes: a first acquisition module 601, a second acquisition module 602, a third acquisition module 603, and an analysis module 604.

[0146] The first acquisition module 601 is used to acquire field patrol task instructions, which include: field patrol time and target farmland location.

[0147] The second acquisition module 602 is used to acquire a first image representing the overall area of ​​the target farmland corresponding to the patrol time using satellite remote sensing technology, based on the location of the target farmland and the patrol time.

[0148] The third acquisition module 603 is used to acquire a second image representing the growth area of ​​the crop to be detected based on the first image using UAV technology, wherein the crop to be detected is the crop in the target farmland;

[0149] The analysis module 604 is used to perform growth risk analysis on the crops to be tested based on the second image, the field inspection time, and a pre-built farmland and crop database, and generate growth risk assessment results for the crops to be tested. The growth risks include: lateral growth risk and longitudinal growth risk. The farmland and crop database includes: farmland location information, crop information associated with the farmland location information, and historical reference images of each growth stage corresponding to the crop information.

[0150] In one example, analysis module 604 includes:

[0151] The retrieval module is used to retrieve reference images from the farmland and crop database based on the location of the target farmland and the information of the crop to be detected. The reference images include: horizontal reference images and vertical reference images. Horizontal reference images refer to historical reference images that are the same as the information of the crop to be detected and whose growth stage matches the field inspection time. Vertical reference images refer to reference images that are located in the same target farmland location as the crop to be detected and whose collection time is within a preset time window before the field inspection time.

[0152] The determination module is used to determine the image feature points of the second image, the horizontal reference image, and the vertical reference image, and extract the feature vectors corresponding to the image feature points;

[0153] The first calculation module is used to calculate, based on a similarity algorithm, a first similarity between the feature vector of the second image and the feature vector of the horizontal reference image, and a second similarity between the feature vector of the second image and the feature vector of the vertical reference image;

[0154] The comparison module is used to compare the first similarity with the horizontal similarity threshold and the second similarity with the vertical similarity threshold to generate a growth risk assessment result.

[0155] In one example, the crop detection device 600 also includes:

[0156] The first extraction module is used to extract a first historical health image dataset from the farmland and crop database that is identical to the information of the crop to be detected and whose growth stage matches the field inspection time.

[0157] The second calculation module is used to calculate the horizontal similarity value between each image in the first historical health image dataset, and to perform statistical distribution analysis on the horizontal similarity value;

[0158] The third calculation module is used to calculate the horizontal similarity threshold based on the statistical distribution of horizontal similarity values.

[0159] In one example, the crop detection device 600 also includes:

[0160] The second extraction module is used to extract, based on each first historical health image in the first historical health image dataset, a second historical health image corresponding to the first historical health image and whose collection time is within a preset time window before it from the farmland and crop database.

[0161] The fourth calculation module is used to calculate the longitudinal similarity value between each first historical health image and its corresponding second historical health image, and to perform statistical distribution analysis on the longitudinal similarity value;

[0162] The fifth calculation module is used to calculate the vertical similarity threshold based on the statistical distribution of vertical similarity values.

[0163] In one example, the second acquisition module 602 includes:

[0164] The fourth acquisition module is used to acquire satellite images corresponding to the patrol time based on the location of the target farmland and the patrol time, and to crop the target farmland image from the satellite images based on the location of the target farmland.

[0165] The third extraction module is used to perform image segmentation on the target farmland image using an edge detection algorithm, extract the edge information of the target farmland, and generate a first image with geographic reference information based on the edge information of the target farmland.

[0166] In one example, the third acquisition module 603 includes:

[0167] The planning module is used to plan the flight path of the UAV based on the geographic reference information recorded in the first image, and control the UAV to take pictures along the flight path to obtain the image sequence of the target farmland.

[0168] The stitching module is used to stitch together the image sequence of the target farmland to obtain UAV images of the target farmland;

[0169] The recognition module is used to process UAV images using image analysis algorithms, identify and segment crop areas, and generate a second image.

[0170] In one example, the crop detection device 600 also includes:

[0171] The fifth acquisition module is used to acquire multiple farmland locations and crop information;

[0172] The first association module is used to associate and map farmland location with crop information to generate structured farmland attribute information;

[0173] The first upload module is used to upload farmland attribute information to the cloud server and save it to the farmland and crop database.

[0174] In one example, the crop detection device 600 also includes:

[0175] The second association module is used to associate and map the target farmland location, the first image, the second image, the field inspection time, and the growth risk assessment results to form a farmland crop data sample.

[0176] The second upload module is used to upload farmland and crop data samples to the cloud server, and based on farmland attribute information, associate and store the farmland and crop data samples in the farmland and crop database.

[0177] The various modules of the crop detection device provided in this application embodiment can achieve Figure 1 It provides the functions for each step of the crop testing method and can achieve the corresponding technical effects. For the sake of brevity, it will not be described in detail here.

[0178] Figure 7 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown.

[0179] An electronic device may include a processor 701 and a memory 702 storing computer program instructions.

[0180] Specifically, the processor 701 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0181] Memory 702 may include mass storage for data or instructions. For example, and not limitingly, memory 702 may include a hard disk drive (HDD), a floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. In one instance, memory 702 may include removable or non-removable (or fixed) media, or memory 702 may be a non-volatile solid-state memory.

[0182] In one instance, memory 702 may be read-only memory (ROM). In one instance, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0183] Memory 702 may include read-only memory (ROM), random access memory (RAM), disk storage media device, optical storage media device, flash memory device, electrical, optical, or other physical / tangible memory storage device. Therefore, generally, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of this disclosure.

[0184] The processor 701 implements a crop detection method in the above-described embodiment by reading and executing computer program instructions stored in the memory 702.

[0185] In one example, the electronic device may also include a communication interface 703 and a bus 704. Wherein, as... Figure 7 As shown, the processor 701, memory 702, and communication interface 703 are connected through bus 704 and complete communication with each other.

[0186] The communication interface 703 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0187] Bus 704 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 704 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0188] The crop detection methods described in the above embodiments can be implemented using a computer storage medium provided in this application. This computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the crop detection methods described in the above embodiments.

[0189] This application also provides a computer program product, including a computer program, which, when executed, implements any of the crop detection methods described in the above embodiments.

[0190] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0191] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0192] The above are merely specific embodiments of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A method of crop detection, characterized by, The method comprises: obtaining a field inspection task instruction, the field inspection task instruction comprising a field inspection time and a target farmland position; based on the target farmland position and the field inspection time, obtaining a first image representing the overall area of the target farmland corresponding to the field inspection time by satellite remote sensing technology; based on the first image, obtaining a second image representing the crop to be detected by using unmanned aerial vehicle technology, the crop to be detected being the crop in the target farmland; based on the second image, the field inspection time and a pre-constructed farmland crop database, performing growth risk analysis on the crop to be detected to generate a growth risk assessment result of the crop to be detected, the growth risk comprising a horizontal growth risk and a vertical growth risk, the farmland crop database comprising farmland position information, crop information associated with the farmland position information, and historical reference images of each growth stage corresponding to the crop information.

2. The method of claim 1, wherein, The method comprises: based on the target farmland position and the crop to be detected information, retrieving a reference image from the farmland crop database, the reference image comprising a horizontal reference image and a vertical reference image, the horizontal reference image being a historical reference image that is the same as the crop to be detected information and whose growth stage matches the field inspection time, and the vertical reference image being a reference image that is located at the same target farmland position as the crop to be detected and whose collection time is within a preset time window before the field inspection time; determining image feature points of the second image, the horizontal reference image and the vertical reference image, and extracting feature vectors corresponding to the image feature points; based on a similarity algorithm, calculating a first similarity between the feature vectors of the second image and the horizontal reference image, and a second similarity between the feature vectors of the second image and the vertical reference image; comparing the first similarity with a horizontal similarity threshold value, and comparing the second similarity with a vertical similarity threshold value to generate a growth risk assessment result.

3. The method of claim 2, wherein, Before comparing the first similarity with the horizontal similarity threshold value, the method further comprises: extracting a first historical health image dataset that is the same as the crop to be detected information and whose growth stage matches the field inspection time from the farmland crop database; calculating horizontal similarity values between images in the first historical health image dataset, and performing statistical distribution analysis on the horizontal similarity values; calculating the horizontal similarity threshold value according to the statistical distribution of the horizontal similarity values.

4. The method of claim 3, wherein, Before comparing the second similarity with the vertical similarity threshold value, the method comprises: extracting, from the farmland crop database, a second historical health image corresponding to each first historical health image in the first historical health image dataset and collected before a preset time window of the first historical health image; calculating a longitudinal similarity value between each first historical health image and the corresponding second historical health image, and performing statistical distribution analysis on the longitudinal similarity values; calculating the longitudinal similarity threshold value according to the statistical distribution of the longitudinal similarity values.

5. The method of claim 1, wherein, The first image representing the overall region of the target farmland corresponding to the time of the field tour is obtained by satellite remote sensing technology based on the target farmland location and the time of the field tour, including: Based on the target farmland location and the time of the field tour, a satellite image corresponding to the time of the field tour is obtained, and a target farmland image is cropped from the satellite image based on the target farmland location; An edge detection algorithm is used to perform image segmentation on the target farmland image to extract edge information of the target farmland, and a first image with geographic reference information is generated based on the edge information of the target farmland.

6. The method of claim 5, wherein, The second image representing the to-be-detected crop is obtained by using unmanned aerial vehicle technology based on the first image, including: According to the geographic reference information recorded in the first image, the flight path of the unmanned aerial vehicle is planned, and the unmanned aerial vehicle is controlled to take pictures along the flight path to obtain a sequence of images of the target farmland; Image stitching is performed on the sequence of images of the target farmland to obtain an unmanned aerial vehicle image of the target farmland; An image analysis algorithm is used to process the unmanned aerial vehicle image to identify and segment the crop area, and generate a second image.

7. The method of claim 1, wherein, Before performing growth risk analysis on the to-be-detected crop based on the second image, the time of the field tour, and the pre-constructed farmland crop database to generate the growth risk assessment result of the to-be-detected crop, it further includes: Obtaining a plurality of farmland locations and crop information; Correlating and mapping the farmland locations and the crop information to generate structured farmland attribute information; Uploading the farmland attribute information to a cloud server and saving it to the farmland crop database.

8. The method of claim 7, wherein, After performing growth risk analysis on the to-be-detected crop based on the second image, the time of the field tour, and the pre-constructed farmland crop database to generate the growth risk assessment result of the to-be-detected crop, it further includes: Correlating and mapping the target farmland location, the first image, the second image, the time of the field tour, and the growth risk assessment result to form a farmland crop data sample; Uploading the farmland crop data sample to the cloud server and storing the farmland crop data sample in association with the farmland attribute information in the farmland crop database.

9. A crop detection apparatus characterized by comprising: The device includes: A first acquisition module for acquiring a field tour task instruction, the field tour task instruction including a field tour time and a target farmland location; A second acquisition module for obtaining a first image representing the overall region of the target farmland corresponding to the time of the field tour by satellite remote sensing technology based on the target farmland location and the time of the field tour; The third acquisition module is configured to acquire, based on the first image, a second image representing a growth area of the crop to be detected by using a UAV technology, the crop to be detected being a crop in a target farmland; The analysis module is configured to perform growth risk analysis on the crop to be detected based on the second image, the time of field inspection, and a pre-constructed farmland crop database, and generate a growth risk assessment result of the crop to be detected, the growth risk including a horizontal growth risk and a vertical growth risk, and the farmland crop database including farmland location information, crop information associated with the farmland location information, and historical reference images of each growth stage corresponding to the crop information.

10. An electronic device, comprising: The device comprises a processor and a memory storing computer program instructions; the processor executes the computer program instructions to implement the crop detection method according to any one of claims 1-8.

11. A computer readable storage medium characterized by, The computer readable storage medium stores computer program instructions, and the computer program instructions are executed by the processor to implement the crop detection method according to any one of claims 1-8.

12. A computer program product, characterised in that, The instructions in the computer program product are executed by the processor of the electronic device, so that the electronic device can execute the crop detection method according to any one of claims 1-8.

Images