Method, device, equipment and medium for preventing and treating crop damage in large plant spacing

Abstract

The present application relates to the field of agricultural technology, and more particularly to a large-plant-spacing crop damage prevention method, device, equipment and medium. The present application realizes accurate pesticide application on the current crop by real-time acquisition of crop images and feature matching to obtain the disease state of the current crop. Compared with the traditional prescription map pesticide application method relying on GNSS for crop positioning, the present application does not introduce positioning errors, and therefore can be applied to large-scale large-plant-spacing crop damage prevention, reducing chemical pesticide waste and improving damage prevention efficiency.

Description

Technical Field

[0001] This invention relates to the field of agricultural technology, and in particular to a method, device, equipment, and medium for preventing and controlling crop damage caused by large plant spacing. Background Technology

[0002] Pests and diseases severely impact crop yield and quality, posing a significant threat to agricultural production. Therefore, preventing and controlling crop damage is a crucial issue currently facing agricultural science.

[0003] In existing technologies, spraying chemical pesticides is the most effective means of pest and disease control. For crops with large plant spacing, a prescription map is usually used to control the pesticide spraying area and ratio, enabling intermittent application of pesticides to both target and non-target areas to reduce pesticide waste in non-target areas. Specifically, a prescription map is first constructed based on information such as the location and severity of damage to the target crop, and then a drone is used to spray the pre-formulated pesticide at the preset locations.

[0004] However, existing technologies have at least the following drawbacks: prescription maps are usually constructed with the planting location as the target, resulting in low accuracy of spraying based on prescription maps and relying on GNSS positioning, which is prone to spraying errors. Summary of the Invention

[0005] This invention provides a method, device, equipment, and medium for the prevention and control of crop damage with large plant spacing, which solves the defects of low application precision and easy spraying error in the prior art, and achieves high-precision chemical pesticide spraying.

[0006] This invention provides a method for controlling crop damage caused by large plant spacing, comprising the following steps:

[0007] Get the current crop image;

[0008] The current crop image is input into the feature extraction network to obtain the current crop feature information output by the feature extraction network;

[0009] The target crop for pesticide application is obtained by matching the current crop characteristic information with the crop characteristic information of the pesticide application prescription map;

[0010] Based on the pesticide application prescription map, query the crop disease status corresponding to the target crop to obtain the current crop disease status;

[0011] Apply pesticides to the current crop based on the current disease status;

[0012] The application prescription map is used to store crop disease status and crop characteristic information. The crop characteristic information stored in the application prescription map is obtained based on a feature extraction network.

[0013] According to the method for controlling crop damage caused by large plant spacing provided by the present invention, the application prescription diagram is obtained according to the following steps:

[0014] Acquire images of individual crop plants;

[0015] Input the single crop image into the disease identification network to obtain the crop disease status output by the disease identification network;

[0016] The single crop image is input into the feature extraction network to obtain the crop feature information output by the feature extraction network;

[0017] Based on the crop disease status and crop characteristic information corresponding to each individual crop image, a pesticide application prescription map is obtained.

[0018] The method for preventing and controlling crop damage caused by large plant spacing according to the present invention includes a disease identification network comprising a disease classification subnetwork and a disease grading subnetwork. The method involves inputting an image of a single crop plant into the disease identification network to obtain the crop disease status output by the network, and includes the following steps:

[0019] Input the single crop image into the disease classification subnetwork to obtain the crop disease type output by the disease classification subnetwork;

[0020] Input the single crop image into the disease classification subnetwork to obtain the crop disease level output by the disease classification subnetwork;

[0021] The crop disease status is determined based on the type and severity of the crop disease.

[0022] According to the method for preventing and controlling crop damage caused by large plant spacing provided by the present invention, the step of acquiring an image of a single crop plant includes the following steps:

[0023] Obtain the image of the first crop plot;

[0024] By stitching together the images of the first crop plot, a global image of the crop plot is obtained;

[0025] The global image of the crop plot is input into the first target detection model to obtain a single crop image output by the first target detection model; the first target detection model is trained on the crop image data through labels.

[0026] According to the method for preventing and controlling crop damage caused by large plant spacing provided by the present invention, the step of acquiring the current crop image includes the following steps:

[0027] Acquire the image of the second crop plot;

[0028] The second crop plot image is input into the second crop target detection model to obtain the current crop image output by the second crop target detection model; the second crop target detection model is trained on crop image data through labels; the structure of the second crop target detection model is the same as that of the first target detection model.

[0029] According to the method for preventing and controlling crop damage caused by large plant spacing provided by the present invention, the step of acquiring the current crop image includes the following steps:

[0030] Acquire the image of the second crop plot;

[0031] The second crop plot image is input into the third crop target detection model to obtain the current crop image and current crop pest status output by the third crop target detection model; the third crop target detection model is trained with labels based on crop image data and pest data; the structure of the third crop target detection model is the same as that of the first target detection model.

[0032] The method for preventing and controlling crop damage caused by large plant spacing also includes the following steps:

[0033] Apply pesticides to the current crop based on the current pest situation.

[0034] This invention also provides a device for controlling crop damage caused by large plant spacing, comprising:

[0035] The image acquisition module is used to acquire the current crop image;

[0036] The feature information module is used to input the current crop image into the feature extraction network and obtain the current crop feature information output by the feature extraction network;

[0037] The crop matching module is used to match the current crop feature information with the crop feature information of the pesticide application prescription map to obtain the target crop for pesticide application;

[0038] The disease query module is used to query the crop disease status corresponding to the target crop based on the pesticide application prescription map, and obtain the current crop disease status.

[0039] The pesticide application module is used to apply pesticides to the current crop based on the current crop disease status.

[0040] The application prescription map is used to store crop disease status and crop characteristic information. The crop characteristic information stored in the application prescription map is obtained based on a feature extraction network.

[0041] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for preventing and controlling crop damage caused by large plant spacing as described above.

[0042] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for preventing and controlling crop damage caused by large plant spacing as described above.

[0043] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the method for preventing and controlling crop damage caused by large plant spacing as described above.

[0044] The present invention provides a method, device, equipment, and medium for the prevention and control of crop damage with large plant spacing. By acquiring crop images in real time and performing feature matching, the current disease status of the crop is obtained, thereby enabling precise application of pesticides to the current crop. Compared with the traditional prescription map application method that relies on GNSS for crop positioning, the present invention does not introduce positioning errors. Therefore, it can be applied to the prevention and control of crop damage with large plant spacing over a wide area, reducing the waste of chemical pesticides and improving the efficiency of damage control. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0046] Figure 1 This is a flowchart illustrating the method for preventing and controlling crop damage caused by large plant spacing provided by the present invention.

[0047] Figure 2 This is a schematic diagram of the feature extraction network structure of the crop damage control method with large plant spacing provided by the present invention;

[0048] Figure 3 This is a schematic diagram of the disease classification subnetwork of the crop damage control method with large plant spacing provided by the present invention;

[0049] Figure 4 This is a schematic diagram of the disease classification subnetwork of the crop damage control method with large plant spacing provided by the present invention;

[0050] Figure 5 This is a schematic diagram of the structure of the first target detection model of the crop damage control method with large plant spacing provided by the present invention;

[0051] Figure 6 This is a schematic diagram of the structure of the large-spacing crop damage control device provided by the present invention;

[0052] Figure 7 This is a schematic diagram of the application module of the crop damage control device with large plant spacing provided by the present invention;

[0053] Figure 8 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0055] The following is combined with Figures 1-8 This invention describes the method, apparatus, equipment, and medium for controlling crop damage caused by large plant spacing.

[0056] Figure 1 This is a flowchart illustrating the method for controlling crop damage caused by large plant spacing according to an embodiment of the present invention, as shown below. Figure 1 As shown, steps 110-150 are included, specifically:

[0057] Step 110: Obtain the current crop image;

[0058] Step 120: Input the current crop image into the feature extraction network to obtain the current crop feature information output by the feature extraction network;

[0059] Step 130: Match the current crop feature information with the crop feature information of the pesticide application prescription map to obtain the target crop for pesticide application;

[0060] Step 140: Query the crop disease status corresponding to the target crop based on the pesticide application prescription map to obtain the current crop disease status;

[0061] Step 150: Apply pesticides to the current crop based on the current crop disease status;

[0062] The application prescription map is used to store crop disease status and crop characteristic information. The crop characteristic information stored in the application prescription map is obtained based on a feature extraction network.

[0063] In step 110 of this embodiment of the invention, the spatial information of the current crop can be determined first, and the space where the current crop is located can be photographed according to the spatial information of the current crop to obtain the current crop image; or a drone can be used to traverse the crop plot, photograph the crop plot to obtain the photographed image, and then the current crop image can be obtained according to the photographed image.

[0064] It is understood that in step 120 of the present invention, since the feature information is only used for feature matching, the feature extraction network can be used to extract arbitrary feature information. For example, in one embodiment of the present invention, the feature extraction network can extract the semantic features of the image based on the current crop image; in another embodiment of the present invention, the feature extraction network can extract the shape and structural features of the image based on the current crop image; it can also extract meaningless features, for example, the feature information extracted by the feature extraction network can be shallow feature information.

[0065] As a specific embodiment of the present invention, the feature information extracted by the feature extraction network is a 256-dimensional feature vector, and the structure of the feature extraction network is as follows: Figure 2 As shown, the model includes a feature extraction layer and a fully connected layer. After the feature extraction layer uses convolution to extract features, the fully connected layer transforms the feature map into a 256-dimensional feature vector. During network training, a fully connected layer and a softmax classifier are added after the model. The training method for classification networks is used, and the model is trained on a crop dataset. The loss function is the cross-entropy loss function, and the optimizer is the stochastic gradient descent (SGD) method.

[0066] Furthermore, in step 130 of this embodiment of the invention, the expression for the feature matching method is as follows:

[0067] M i =(y1,y2,y3,…,y 255 ,y 256 )

[0068] D = (x1, x2, x3, ..., x 255 ,x 256 )

[0069]

[0070] num = min(d1, d2, ..., d) n )

[0071] Among them, M i d(M) is the feature vector stored in the pesticide application prescription map, D is the current crop feature vector, and d(M) is the feature vector stored in the map. i D i ) represents the distance between two vectors, and num indicates that the feature vector with the closest distance is taken as the best match result.

[0072] It is understood that in other embodiments of the present invention, the feature vector can be a vector of other dimensions or other data forms, and the matching method can also adopt any similarity measurement method to achieve the purpose of the present invention.

[0073] In this embodiment of the invention, for crops that have been successfully matched and have completed application, the corresponding data can be extracted from the application prescription map so that the information of crops that have completed application will no longer participate in the matching, thus avoiding repeated application of the same crop; at the same time, it can also ensure that all crops have completed application without omission.

[0074] Therefore, the crop damage control method with large plant spacing provided in this embodiment of the invention obtains the current crop disease status by acquiring crop images in real time and performing feature matching, thereby achieving precise application of pesticides to the current crop. Compared with the traditional prescription map application method that relies on GNSS for crop positioning, the method of this embodiment of the invention does not introduce positioning errors. Therefore, it can be applied to the control of crop damage with large plant spacing over a wide area, reducing the waste of chemical pesticides and improving the efficiency of damage control.

[0075] In this embodiment of the invention, the drug application prescription diagram is obtained according to the following steps:

[0076] Acquire images of individual crop plants;

[0077] Input the single crop image into the disease identification network to obtain the crop disease status output by the disease identification network;

[0078] The single crop image is input into the feature extraction network to obtain the crop feature information output by the feature extraction network;

[0079] Based on the crop disease status and crop characteristic information corresponding to each individual crop image, a pesticide application prescription map is obtained.

[0080] The single crop image in this embodiment of the invention can be acquired individually or obtained by segmenting the overall work area image; to ensure pesticide coverage, the single crop image in this embodiment of the invention should include crop images of all crop plants in the work area; the single crop image in this embodiment of the invention is an image of a single crop, which can be a single image or a multi-angle image; each single crop image in this embodiment of the invention is assigned a unique identification tag to the single crop.

[0081] The embodiments of the present invention obtain a drug application prescription map through the above steps, and distinguish the drug application target by means of features and identification tags. Compared with the prescription map technology of the prior art that uses GNSS method to determine the drug application target, the drug application prescription map of the embodiments of the present invention does not introduce positioning error, thereby further improving the drug application accuracy.

[0082] In this embodiment of the invention, the disease identification network includes a disease classification subnetwork and a disease grading subnetwork. Inputting the single crop image into the disease identification network to obtain the crop disease status output by the network includes the following steps:

[0083] Input the single crop image into the disease classification subnetwork to obtain the crop disease type output by the disease classification subnetwork;

[0084] Input the single crop image into the disease classification subnetwork to obtain the crop disease level output by the disease classification subnetwork;

[0085] The crop disease status is determined based on the type and severity of the crop disease.

[0086] Specifically, the structural diagram of the disease classification sub-network provided in the embodiments of the present invention is as follows: Figure 3 As shown, the disease classification sub-network of this embodiment is used to classify the disease types of a single crop. Different classification categories should be set for different disease types. It can be understood that one classification category can correspond to one disease or multiple diseases. In this embodiment, the disease types are divided into healthy, disease I, disease II, and disease III. The backbone feature extraction network consists of PConv and Swin Transformer modules. The dual backbone network is more conducive to the extraction of effective feature information. The features extracted by the two backbone networks are fused by weight addition to further increase the expressive power of effective features. Finally, the sofmax classifier is used to classify the features and output the crop disease type.

[0087] The structural diagram of the disease classification sub-network provided in this embodiment of the invention is as follows: Figure 4 As shown, the disease grading sub-network of this embodiment is used to grade the severity of diseases. Different grading levels can be selected based on actual applications, such as disease type and application precision. In this embodiment, the disease level is divided into three levels: healthy, moderate, and severe. Classifying the severity of diseases is often more difficult than classifying disease types, requiring the extraction of more accurate feature information. Similarly, a dual-backbone feature extraction network is used to extract shallow features. Compact Bilinear Pooling (CBP) is used to fuse the feature maps. After fusion, the width and height of the feature map are halved, while the number of channels is doubled. This gradually increases the receptive field size, thus helping to better extract features with global information. Deep image features contain richer semantic information. A parallel three-feature extraction network is used to obtain deep semantic features for disease grading. CBP is also used for multi-feature fusion. Finally, a Sofmax classifier is used to classify the features and output the crop disease level.

[0088] The embodiments of the present invention achieve automatic generation of pesticide application prescription maps with large-scale coverage and wide plant spacing through the above steps, providing information on crop disease types and disease levels, providing data support for pesticide application formulation decisions, and improving the efficiency of pesticide application to crops.

[0089] In this embodiment of the invention, acquiring a single crop image includes the following steps:

[0090] Obtain the image of the first crop plot;

[0091] By stitching together the images of the first crop plot, a global image of the crop plot is obtained;

[0092] The global image of the crop plot is input into the first target detection model to obtain a single crop image output by the first target detection model; the first target detection model is trained on the crop image data through labels.

[0093] Specifically, an orthophoto image of the crop plot was obtained by using a drone to collect the crop plot image. The drone flew at an altitude of 1.20m. The acquired images were stitched together using Agisoft Metashape Pro image processing software to obtain an aerial image of the entire crop plot. The acquired crop plot image matrix was filled with 0 so that the length and width of the filled image were both multiples of 32.

[0094] The structure of the first target detection model in this embodiment of the invention is as follows: Figure 5 As shown, the first target detection model in this embodiment of the invention is based on the YOLO v5 framework. The backbone network extracts crop features from the plot image for subsequent target detection. Partial Convolution (PConv) is located in layers 1, 2, 4, 6, and 8 of the model. PConv can effectively reduce model parameters and improve image processing efficiency. Layers 3, 5, 7, and 9 are C3 structures, and layer 10 uses an SPFF structure to fuse features. The Neck part of the network adopts a PANet structure to generate a feature pyramid. The feature pyramid enhances the model's detection of crops at different scaling scales, thereby enabling the identification of the same crop individual of different sizes and scales, further improving the model's recognition accuracy. The upper part uses a Content-Aware ReAssembly of Features (CARAFE) upsampling method to increase the receptive field and aggregate contextual information in the crop image, making the upsampled feature map information more complete. Based on the recognition results of the first target detection model, the global image of the crop plot is segmented to obtain individual crop images.

[0095] It is understood that, in embodiments of the present invention, drones can be used to take multiple photos of the same plot of land from multiple different angles, and the same rules can be used to encode the recognition results of the first target detection model. For example, the encoding can be done sequentially from top to bottom and from left to right, so that the encoding corresponds one-to-one with individual crop plants, thereby obtaining multi-angle images of single crops.

[0096] In this embodiment of the invention, a first target detection model is used to detect the global image of crop plots with large plant spacing and large area, and the model is optimized for the identification of the same crop individual of different sizes and scales to improve the identification accuracy and thus improve the generation efficiency of pesticide prescription map.

[0097] In this embodiment of the invention, acquiring the current crop image includes the following steps:

[0098] Acquire the image of the second crop plot;

[0099] The second crop plot image is input into the second crop target detection model to obtain the current crop image output by the second crop target detection model; the second crop target detection model is trained on crop image data through labels; the structure of the second crop target detection model is the same as that of the first target detection model.

[0100] In this embodiment of the invention, a drone crop spraying device traverses crop plots, takes pictures of the crop plots to obtain a second crop plot image, and then obtains the current crop image based on the second crop plot image. Since the second crop target detection model and the first target detection model use the same structure, the recognition results for the same type of crop are similar, reducing the false recognition rate of crops based on the second crop target detection model and improving recognition efficiency.

[0101] In another embodiment of the present invention, acquiring the current crop image includes the following steps:

[0102] Acquire the image of the second crop plot;

[0103] The second crop plot image is input into the third crop target detection model to obtain the current crop image and current crop pest status output by the third crop target detection model; the third crop target detection model is trained with labels based on crop image data and pest data; the structure of the third crop target detection model is the same as that of the first target detection model.

[0104] The method for preventing and controlling crop damage caused by large plant spacing also includes the following steps:

[0105] Apply pesticides to the current crop based on the current pest situation.

[0106] In this embodiment of the invention, the crop spraying device of the drone traverses the crop plot, and while identifying and matching characteristics, it can also identify pests and apply pesticides accordingly, so as to achieve integrated prevention and control of diseases and pests and improve the efficiency of pesticide application.

[0107] In summary, the crop damage control method with large plant spacing provided by this invention obtains the current crop disease status by real-time acquisition of crop images and feature matching, thereby achieving precise application of pesticides to the current crop. Compared with the traditional prescription map application method that relies on GNSS for crop positioning, this invention does not introduce positioning errors. Therefore, it can be applied to the control of crop damage over a large area with large plant spacing, reducing the waste of chemical pesticides and improving the efficiency of damage control.

[0108] The following describes the large-spacing crop damage control device provided by the present invention. The large-spacing crop damage control device described below and the large-spacing crop damage control method described above can be referred to in correspondence.

[0109] Figure 6 This is a schematic diagram of the structure of the large-spacing crop damage control device provided by the present invention, as shown below. Figure 6 As shown, it includes:

[0110] Image acquisition module 610 is used to acquire the current crop image;

[0111] The feature information module 620 is used to input the current crop image into the feature extraction network and obtain the current crop feature information output by the feature extraction network;

[0112] The crop matching module 630 is used to match the current crop feature information with the crop feature information of the pesticide application prescription map to obtain the target crop for pesticide application;

[0113] The disease query module 640 is used to query the crop disease status corresponding to the target crop based on the pesticide application prescription map, and obtain the current crop disease status.

[0114] The pesticide application module 650 is used to apply pesticides to the current crop based on the current crop disease status;

[0115] The application prescription map is used to store crop disease status and crop characteristic information. The crop characteristic information stored in the application prescription map is obtained based on a feature extraction network.

[0116] Therefore, the crop damage control device with large plant spacing provided by the present invention can obtain the current crop disease status by acquiring crop images in real time and performing feature matching, thereby achieving precise application of pesticides to the current crop. Compared with the traditional prescription map application method that relies on GNSS for crop positioning, the present invention does not introduce positioning errors. Therefore, it can be applied to the prevention and control of crop damage with large plant spacing, reducing the waste of chemical pesticides and improving the efficiency of damage control.

[0117] In this embodiment of the invention, the application module 650 can be a spray boom device equipped with a multi-way solenoid valve, such as... Figure 7As shown, during spraying, the spray boom equipped with multiple solenoid valves is positioned directly above the crop. The application module controls the opening and closing of different solenoid valves based on the disease information in the prescription map and the location and size of the crop, and adjusts the duty cycle of the solenoid valves according to the severity of the disease to control the amount of pesticide applied. Based on real-time identification of pest types and locations, the pest control system achieves precise variable-rate pesticide application to the crop.

[0118] Figure 8 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 8 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute a method for preventing and controlling damage to crops with large plant spacing, including the following steps:

[0119] Get the current crop image;

[0120] The current crop image is input into the feature extraction network to obtain the current crop feature information output by the feature extraction network;

[0121] The target crop for pesticide application is obtained by matching the current crop characteristic information with the crop characteristic information of the pesticide application prescription map;

[0122] Based on the pesticide application prescription map, query the crop disease status corresponding to the target crop to obtain the current crop disease status;

[0123] Apply pesticides to the current crop based on the current disease status;

[0124] The application prescription map is used to store crop disease status and crop characteristic information. The crop characteristic information stored in the application prescription map is obtained based on a feature extraction network.

[0125] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0126] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the crop damage control methods for large plant spacing provided by the above methods, including the following steps:

[0127] Get the current crop image;

[0128] The current crop image is input into the feature extraction network to obtain the current crop feature information output by the feature extraction network;

[0129] The target crop for pesticide application is obtained by matching the current crop characteristic information with the crop characteristic information of the pesticide application prescription map;

[0130] Based on the pesticide application prescription map, query the crop disease status corresponding to the target crop to obtain the current crop disease status;

[0131] Apply pesticides to the current crop based on the current disease status;

[0132] The application prescription map is used to store crop disease status and crop characteristic information. The crop characteristic information stored in the application prescription map is obtained based on a feature extraction network.

[0133] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the methods for preventing and controlling crop damage caused by large plant spacing provided by the above methods, including the following steps:

[0134] Get the current crop image;

[0135] The current crop image is input into the feature extraction network to obtain the current crop feature information output by the feature extraction network;

[0136] The target crop for pesticide application is obtained by matching the current crop characteristic information with the crop characteristic information of the pesticide application prescription map;

[0137] Based on the pesticide application prescription map, query the crop disease status corresponding to the target crop to obtain the current crop disease status;

[0138] Apply pesticides to the current crop based on the current disease status;

[0139] The application prescription map is used to store crop disease status and crop characteristic information. The crop characteristic information stored in the application prescription map is obtained based on a feature extraction network.

[0140] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0141] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0142] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for controlling crop damage caused by large plant spacing, characterized in that, Includes the following steps: Get the current crop image; The current crop image is input into the feature extraction network to obtain the current crop feature information output by the feature extraction network; The target crop for pesticide application is obtained by matching the current crop characteristic information with the crop characteristic information of the pesticide application prescription map; Based on the pesticide application prescription map, query the crop disease status corresponding to the target crop to obtain the current crop disease status; Apply pesticides to the current crop based on the current disease status; The application prescription map is used to store crop disease status and crop characteristic information. The crop characteristic information stored in the application prescription map is obtained based on a feature extraction network. The drug application prescription diagram was obtained according to the following steps: Acquire images of individual crop plants; Input the single crop image into the disease identification network to obtain the crop disease status output by the disease identification network; The single crop image is input into the feature extraction network to obtain the crop feature information output by the feature extraction network; Based on the crop disease status and crop characteristic information corresponding to each individual crop image, a pesticide application prescription map is obtained.

2. The method for preventing and controlling crop damage caused by large plant spacing according to claim 1, characterized in that, The disease identification network includes a disease classification subnetwork and a disease grading subnetwork. Inputting the single crop image into the disease identification network and obtaining the crop disease status output by the network includes the following steps: Input the single crop image into the disease classification subnetwork to obtain the crop disease type output by the disease classification subnetwork; Input the single crop image into the disease classification subnetwork to obtain the crop disease level output by the disease classification subnetwork; The crop disease status is determined based on the type and severity of the crop disease.

3. The method for preventing and controlling crop damage caused by large plant spacing according to claim 1, characterized in that, The process of acquiring images of individual crops includes the following steps: Obtain the image of the first crop plot; By stitching together the images of the first crop plot, a global image of the crop plot is obtained; The global image of the crop plot is input into the first target detection model to obtain a single crop image output by the first target detection model; the first target detection model is trained on the crop image data through labels.

4. The method for preventing and controlling crop damage caused by large plant spacing according to claim 3, characterized in that, The process of acquiring the current crop image includes the following steps: Acquire the image of the second crop plot; The second crop plot image is input into the second crop target detection model to obtain the current crop image output by the second crop target detection model; the second crop target detection model is trained on crop image data through labels; the structure of the second crop target detection model is the same as that of the first target detection model.

5. The method for preventing and controlling crop damage caused by large plant spacing according to claim 3, characterized in that, The process of acquiring the current crop image includes the following steps: Acquire the image of the second crop plot; The second crop plot image is input into the third crop target detection model to obtain the current crop image and current crop pest status output by the third crop target detection model; the third crop target detection model is trained with labels based on crop image data and pest data; the structure of the third crop target detection model is the same as that of the first target detection model. The method for preventing and controlling crop damage caused by large plant spacing also includes the following steps: Apply pesticides to the current crop based on the current pest situation.

6. A device for preventing and controlling crop damage with large plant spacing, characterized in that, include: The image acquisition module is used to acquire the current crop image; The feature information module is used to input the current crop image into the feature extraction network and obtain the current crop feature information output by the feature extraction network; The crop matching module is used to match the current crop feature information with the crop feature information of the pesticide application prescription map to obtain the target crop for pesticide application; The disease query module is used to query the crop disease status corresponding to the target crop based on the pesticide application prescription map, and obtain the current crop disease status. The pesticide application module is used to apply pesticides to the current crop based on the current crop disease status. The application prescription map is used to store crop disease status and crop characteristic information. The crop characteristic information stored in the application prescription map is obtained based on a feature extraction network. The drug application prescription diagram was obtained according to the following steps: Acquire images of individual crop plants; Input the single crop image into the disease identification network to obtain the crop disease status output by the disease identification network; The single crop image is input into the feature extraction network to obtain the crop feature information output by the feature extraction network; Based on the crop disease status and crop characteristic information corresponding to each individual crop image, a pesticide application prescription map is obtained.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for preventing and controlling crop damage caused by large plant spacing as described in any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for preventing and controlling crop damage caused by large plant spacing as described in any one of claims 1 to 5.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for preventing and controlling crop damage caused by large plant spacing as described in any one of claims 1 to 5.

Images