Seedling missing rate calculation method, system and equipment, medium and product

Full-supervised and semi-supervised training was carried out through the neural network model, combined with field ridge line extraction, and the problem of low-efficiency calculation efficiency in crop fields was solved, achieving rapid and accurate seedless rate detection.

CN118982771AActive Publication Date: 2024-11-19ZHEJIANG UNIV +1
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202410951580.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-16
Publication Date
2024-11-19
Estimated Expiration
2044-07-16

AI Technical Summary

Technical Problem

In the prior art, the calculation efficiency of crop field seedling shortage rate is low, mainly due to the small seedling size and the drone remote sensing technology requires a large amount of data, which makes the labeling and data sets time-consuming and labor-intensive, and the calculation efficiency is not high.

Method used

The neural network model is used for full supervision and semi-supervised training. The sample image is detected through the seedless area detection model, combined with the field ridge line extraction to calculate the seedless rate, and the mapping relationship and prediction frame are used to reduce the image processing calculation amount to achieve rapid calculate.

Benefits of technology

This greatly reduces the workload and time of the seedling shortage rate detection, improves the calculation efficiency, and can quickly and accurately obtain the seedling shortage rate of the land to be tested.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118982771A_ABST
    Figure CN118982771A_ABST
Patent Text Reader

Abstract

The invention discloses a seedling missing rate calculation method, system and device, a medium and a product, and relates to the field of crop seedling missing detection, and the method comprises the steps: obtaining a pre-shot sample image of crops of a to-be-detected land parcel and a mapping relation; carrying out seedling-lacking area detection on the sample image by utilizing the seedling-lacking area detection model to obtain a prediction frame of the seedling-lacking area; using the mapping relation and the prediction frame to obtain a bounding box of a seedling missing area in the original image; performing field ridge line extraction on the original image to obtain a vertical field ridge line of the to-be-detected land parcel; obtaining the seedling missing rate of the original image based on the vertical field ridge line and the bounding box; and obtaining an average value of the seedling missing rates by using the seedling missing rates of the plurality of original images, and taking the average value as the seedling missing rate of the to-be-detected land parcel. According to the method, the calculation efficiency of the seedling missing rate of large-area planted crops is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of crop seedling deficiency detection, and in particular to a method, system, equipment, medium and product for calculating a seedling deficiency rate. Background Art

[0002] The germination of seedlings during the seedling stage of crops is of great guiding significance for field management, and the seedling deficiency rate is one of the evaluation indicators of seedling germination. The traditional field seedling deficiency rate relies on manual field inspection and visual estimation, which is often not accurate enough.

[0003] With the development of UAV remote sensing technology, the calculation of crop field seedling missing rate is more efficient and accurate when combined with UAV remote sensing technology and computer vision technology. However, since the seedlings of crops are small in size during the seedling stage, and UAV remote sensing technology requires a large amount of data, it is time-consuming and laborious to mark the seedling missing area and produce data sets. Therefore, the calculation of seedling missing rate in the existing technology is time-consuming and labor-intensive, and has low calculation efficiency. Summary of the invention

[0004] The purpose of this application is to provide a method, system, equipment, medium and product for calculating the seedling deficiency rate, which can improve the calculation efficiency of the seedling deficiency rate of crops planted over a large area.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a method for calculating the seedling deficiency rate, comprising:

[0007] Obtaining a sample image of crops of a plot to be inspected that is pre-photographed and a mapping relationship; wherein the mapping relationship is a positional relationship between the sample image and an original image; and the original image contains a plurality of sample images;

[0008] The missing seedling area detection model is used to detect the missing seedling area of ​​the sample image to obtain a prediction frame of the missing seedling area; wherein the missing seedling area detection model is obtained by performing full supervision and semi-supervised training on a neural network model;

[0009] Using the mapping relationship and the prediction frame, a boundary frame of the seedling-missing region in the original image is obtained;

[0010] Extracting ridge lines from the original image to obtain vertical ridge lines of the plot to be detected;

[0011] Based on the vertical ridge line and the boundary box, obtaining the seedling missing rate of the original image;

[0012] The seedling absence rates of a plurality of the original images are used to obtain an average value of the seedling absence rates, and the average value is used as the seedling absence rate of the plot to be detected.

[0013] Furthermore, before obtaining the sample images of the crops of the plot to be detected and the mapping relationship taken in advance, the method further includes:

[0014] Acquire the original image and construct a rectangular coordinate system; wherein the origin of the rectangular coordinate system is the point at the upper left corner of the original image;

[0015] Cut the original images in order without overlapping to obtain a plurality of sample images, and record the cut coordinates of the upper left corner of the sample image in the rectangular coordinate system;

[0016] The mapping relationship is obtained by using the clipping coordinates.

[0017] Furthermore, the obtaining of the original image specifically includes:

[0018] Obtain color images taken by the drone when it flies along the vertical ridge line;

[0019] The edge cutting is performed based on the center point of the color image to obtain a plurality of the original images.

[0020] Further, ridge lines are extracted from the original image to obtain vertical ridge lines of the plot to be detected, specifically including:

[0021] The original image is sequentially subjected to graying, Gaussian filtering, edge detection, vertical line detection based on Hough transform and morphological operation to extract ridge lines, thereby obtaining the vertical ridge lines of the plot to be detected.

[0022] Further, using the vertical ridge line and the boundary box, the missing seedling rate of the original image is obtained, including:

[0023] Calculating the total length of the vertical ridge lines and the sum of the heights of the boundary box;

[0024] The ratio of the total height to the total length is taken as the missing seedling rate of the original image.

[0025] Furthermore, the training process of the seedling-deficient area detection model is as follows:

[0026] Constructing a data set, the data set includes a labeled sample image and an unlabeled sample image, the label information of the labeled sample image includes a first bounding box and a category of the seedling-missing area, wherein the category is the seedling-missing area and the background area;

[0027] Input the labeled sample image into the neural network model to obtain a first prediction box of the missing seedling area, calculate a first loss function according to the first prediction box and the first bounding box, update the parameters of the neural network model according to the first loss function, obtain a fully supervised target detection model, and use the fully supervised target detection model as a teacher model and a student model;

[0028] Input the unlabeled sample image into the teacher model to obtain a sample image with a pseudo label, input the sample image with the pseudo label into the student model to obtain a second prediction box of the seedling-missing area; wherein the pseudo label includes a second bounding box marking the seedling-missing area;

[0029] A second loss function is calculated according to the second prediction box and the second bounding box, and the parameters of the student model are updated according to the second loss function to obtain a missing seedling area detection model.

[0030] In a second aspect, the present application provides a system for calculating the seedling shortage rate, the system for calculating the seedling shortage rate comprising:

[0031] An acquisition module, used to acquire a sample image of crops of a plot to be detected that has been taken in advance and a mapping relationship; wherein the mapping relationship is a positional relationship between the sample image and an original image; and the original image contains a plurality of sample images;

[0032] A prediction module is used to perform seedling-missing region detection on the sample image using a seedling-missing region detection model to obtain a prediction frame of the seedling-missing region; wherein the seedling-missing region detection model is obtained by performing full-supervision and semi-supervised training on a neural network model;

[0033] A seedling-missing region determination module, used to obtain a boundary frame of the seedling-missing region in the original image by using the mapping relationship and the prediction frame;

[0034] A ridge line extraction module is used to extract ridge lines from the original image to obtain vertical ridge lines of the plot to be detected;

[0035] A first calculation module is used to obtain the seedling absence rate of the original image based on the vertical field ridge line and the boundary box;

[0036] The second calculation module is used to use the seedling absence rates of multiple original images to obtain an average value of the seedling absence rates, and use the average value as the seedling absence rate of the plot to be detected.

[0037] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-mentioned method for calculating the seedling absence rate.

[0038] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned method for calculating the seedling deficiency rate.

[0039] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for calculating the seedling deficiency rate.

[0040] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0041] The present application provides a method, system, device, medium and product for calculating the missing seedling rate. The missing seedling area detection model is used to detect the missing seedling area without a large amount of labeled data, thereby greatly reducing the workload and improving the detection efficiency of the missing seedling rate. The prediction box and the mapping relationship are used to obtain the bounding box of the missing seedling area in the original image. The mapping relationship between the original image and the sample image position is used, and the image splicing is not required, which can reduce the amount of image processing calculation and further improve the calculation efficiency of the missing seedling rate. Finally, the missing seedling rate of the original image is calculated using the bounding box and the vertical ridge line; the average value of the missing seedling rate is calculated based on the missing seedling rates of multiple original images, and the average value is used as the missing seedling rate of the plot to be detected, so as to quickly obtain the missing seedling rate of the plot to be detected, thereby greatly improving the calculation efficiency of the missing seedling rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0043] Figure 1 A schematic diagram of a flow chart of a method for calculating a missing seedling rate provided in an embodiment of the present application;

[0044] Figure 2 for Figure 1 A detailed flowchart of the steps for obtaining the mapping relationship;

[0045] Figure 3 A schematic diagram of the functional modules of a system for calculating the seedling deficiency rate provided in one embodiment of the present application. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0047] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0048] In an exemplary embodiment, Figure 1 As shown, a method for calculating the seedling shortage rate is provided, comprising the following steps 101 to 106. Among them:

[0049] Step 101, obtaining a sample image of crops of a plot to be inspected that is pre-photographed and a mapping relationship; wherein the mapping relationship is a positional relationship between the sample image and an original image; and the original image includes a plurality of sample images.

[0050] As an optional implementation, Figure 2 As shown, before the above step 101, the method further includes the following steps 201 to 203. Among them:

[0051] Step 201, obtain the original image and construct a rectangular coordinate system; wherein the origin of the rectangular coordinate system is the point at the upper left corner of the original image.

[0052] On the original image, the point at the upper left corner of the original image is taken as the origin of the rectangular coordinate system, the right is the positive direction of the x-axis, and the downward is the positive direction of the y-axis.

[0053] As an optional implementation, the obtaining step 201 includes the following steps:

[0054] Step 2011, obtaining a color image taken by the drone when flying along the vertical field ridge line.

[0055] The heading repetition rate of the drone was set to 60%, the sideways repetition rate of the drone was set to 60%, the flight speed was 2.5 m / s, and the flight altitude was 25 m. The DJI Mavic3E drone was used to collect color images of the corn at the second leaf stage of the plot to be tested along the ridge direction, with a pixel resolution of 5280×3956.

[0056] Step 2012, edge cropping is performed based on the center point of the color image to obtain a plurality of original images.

[0057] The obtained high-resolution original image, ie, the above-mentioned image with a pixel resolution of 5280×3956, is cropped along the center point to obtain an original image with a pixel resolution of 5120×2560.

[0058] For a large-scale crop planting base, the drone needs to take multiple color images to cover the entire crop planting base. Therefore, after processing in step 3012, multiple original images with a pixel resolution of 5120×2560 can be obtained.

[0059] Step 202 , cropping the original image in sequence without overlapping to obtain a plurality of sample images, and recording the cropping coordinates of the point at the upper left corner of the sample image in the rectangular coordinate system.

[0060] In this implementation, the original image obtained in step 2011 is cropped, and an original image is cropped into a plurality of sample images with a pixel resolution of 640×640 in sequence without overlap, and each sample image is recorded with its corresponding cropping coordinates. In this embodiment, 32 sample images can be obtained after cropping an original image.

[0061] Step 203: Obtain the mapping relationship using the clipping coordinates.

[0062] After processing the multiple original images in step 202, multiple clipping coordinates are obtained. These clipping coordinates constitute the positional relationship between the original image and the sample image in the rectangular coordinate system, that is, the mapping relationship.

[0063] Step 102, using a missing seedling area detection model to perform missing seedling area detection on the sample image to obtain a prediction frame of the missing seedling area; wherein the missing seedling area detection model is obtained by performing full supervision and semi-supervised training on a neural network model.

[0064] The training of the missing seedling area detection model is divided into two stages: the first stage is to use labeled sample images to perform fully supervised training on the neural network model to obtain a fully supervised target detection model; the second stage is to use unlabeled sample images to perform semi-supervised training on the fully supervised neural network model obtained in the first stage under a semi-supervised training framework to obtain a missing seedling area detection model.

[0065] Exemplarily, the neural network model can be a YOLOv8 model or other neural network models. This embodiment is specifically described using the YOLOv8 model. Label information is added to the collected sample images to construct a data set, the data set includes the sample images and label information of the sample images, the label information includes the first bounding box and category of the seedling-missing area, wherein the category includes the seedling-missing area and the background area, the data set is divided into a training set, a validation set and a test set according to a certain ratio (such as 7:1:2), and the training set is divided into at least 8 batches.

[0066] Set hyperparameters (such as learning rate, loss function, etc.) and iterate training for no less than 300 batches so that the prediction results of the YOLOv8 model continuously fit the label information. Perform performance verification on the validation set after each batch of training until the indicators on the validation set tend to be flat, indicating that the model has been fully fitted. Save the trained YOLOv8 model parameters to obtain a fully supervised target detection model for seedling-deficient areas.

[0067] Exemplarily, the second stage uses a teacher-student semi-supervised training framework to perform semi-supervised training on the fully supervised target detection model obtained in the first stage. The YOLOv8 model that has undergone full supervision training, that is, the fully supervised target detection model, is used as the initialized student model and teacher model. During the training process, the unlabeled sample image is input into the teacher model to generate a sample image with a pseudo label, wherein the pseudo label includes a second bounding box marking the seedling-missing area; the sample image with the pseudo label is input into the student model to obtain the second prediction box of the seedling-missing area, and the loss function is calculated based on the second prediction box and the second bounding box, so that the student model outputs the same enhanced versions of the same image during the training process, thereby updating the parameters of the student model. The updated student model optimizes and updates the teacher model through the exponential sliding average method, so that the teacher model generates a better new round of pseudo labels based on the unlabeled sample images, thereby completing a training iteration.

[0068] By setting the learning rate and training batches, after multiple iterations of training, the indicators of the YOLOv8 model on the validation set tend to be stable, thus obtaining a missing seedling area detection model for missing seedling area detection. In this way, a large amount of unlabeled data is used to enhance the performance of missing seedling area target detection, where the ratio of unlabeled data to fully labeled data is greater than 2:1.

[0069] Step 103, using the mapping relationship and the prediction box, obtain a boundary box of the seedling-missing area in the original image.

[0070] The prediction frame of the missing seedling area detected in step 102 is mapped back to the original image according to the clipping coordinates recorded in the mapping relationship to obtain the original image with the prediction frame of the missing seedling area, that is, the boundary frame of the missing seedling area in the original image.

[0071] Step 104, extracting the ridge lines of the original image to obtain the vertical ridge lines of the plot to be detected.

[0072] As an optional implementation, the original image is subjected to grayscale conversion, Gaussian filtering, edge detection, vertical line detection based on Hough transform, and morphological operations in sequence to extract ridge lines, thereby obtaining the vertical ridge lines of the plot to be detected.

[0073] Step 105: obtaining the seedling absence rate of the original image based on the vertical ridge lines and the boundary box.

[0074] As an optional implementation, the total length of the vertical ridge line and the sum of the heights of the boundary box are calculated;

[0075] The ratio of the total height to the total length is taken as the missing seedling rate of the original image.

[0076] In this embodiment, corn is taken as an example of crops. Generally, large-scale planting adopts 2 rows of corn or 4 rows of corn to be planted in the same ridge. The area between two vertical ridge lines is a ridge. In this embodiment, the missing seedling rate is calculated by taking 2 rows of corn planted between ridges as an example. If 4 rows of corn are planted between ridges, the calculation of the missing seedling rate of the original image is adjusted by a corresponding multiple. For example, if 4 rows of corn are planted between ridges, half of the ratio of the total height of the boundary box to the total length of the vertical ridge lines is used as the missing seedling rate of the original image.

[0077] Step 106, using the seedling absence rates of the plurality of original images, obtain an average value of the seedling absence rates, and use the average value as the seedling absence rate of the plot to be detected.

[0078] Implement the above-mentioned steps 101 to 106, by obtaining the sample images of crops of the plot to be detected that have been pre-photographed and the mapping relationship, use the missing seedling area detection model to predict the missing seedling area to obtain a prediction frame, and use the missing seedling area detection model to detect the missing seedling area without a large amount of labeled data, thereby greatly reducing the workload and improving the detection efficiency. At the same time, the missing seedling area detection model adopts a semi-supervised target detection model to improve the accuracy of detecting the missing seedling rate, and uses the prediction frame and the mapping relationship to obtain the bounding box of the missing seedling area in the original image. Using the mapping relationship between the original image and the sample image position, there is no need to splice the image, which can reduce the amount of image processing calculations and further improve the calculation efficiency of the missing seedling rate. Finally, use the bounding box and the vertical field ridge line to calculate the missing seedling rate of the original image; calculate the average value of the missing seedling rate based on the missing seedling rates of multiple original images, and use the average value as the missing seedling rate of the plot to be detected, so as to quickly obtain the missing seedling rate of the plot to be detected, thereby greatly improving the calculation efficiency of the missing seedling rate.

[0079] Based on the same inventive concept, the embodiment of the present application further provides a system for implementing the above-mentioned calculation of the missing seedling rate. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in the one or more embodiments of the calculation system for the missing seedling rate provided below can refer to the limitations of the calculation method for the missing seedling rate in the above text, and will not be repeated here.

[0080] In an exemplary embodiment, Figure 3 As shown, a system for calculating the seedling shortage rate is provided, comprising:

[0081] The acquisition module 31 is used to acquire a sample image of crops of a plot to be detected that is pre-photographed and a mapping relationship; wherein the mapping relationship is a positional relationship between the sample image and an original image; and the original image includes a plurality of sample images.

[0082] The prediction module 32 is used to perform seedling-missing area detection on the sample image using a seedling-missing area detection model to obtain a prediction frame of the seedling-missing area; wherein the seedling-missing area detection model is obtained by performing full supervision and semi-supervised training on a neural network model.

[0083] The seedling-missing area determining module 33 is used to obtain the boundary box of the seedling-missing area in the original image by using the mapping relationship and the prediction box.

[0084] The ridge line extraction module 34 is used to extract the ridge lines from the original image to obtain the vertical ridge lines of the plot to be detected.

[0085] The first calculation module 35 is used to obtain the seedling absence rate of the original image based on the vertical ridge line and the boundary box.

[0086] The second calculation module 36 is used to use the seedling absence rates of multiple original images to obtain an average value of the seedling absence rates, and use the average value as the seedling absence rate of the plot to be detected.

[0087] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0088] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0089] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0090] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0091] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0092] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0093] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0094] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for calculating the seedling shortage rate, characterized in that: The calculation method of the seedling shortage rate includes: Obtaining a sample image of crops of a plot to be inspected that is pre-photographed and a mapping relationship; wherein the mapping relationship is a positional relationship between the sample image and an original image; and the original image contains a plurality of sample images; The missing seedling area detection model is used to detect the missing seedling area of ​​the sample image to obtain a prediction frame of the missing seedling area; wherein the missing seedling area detection model is obtained by performing full supervision and semi-supervised training on a neural network model; Using the mapping relationship and the prediction frame, a boundary frame of the seedling-missing region in the original image is obtained; Extracting ridge lines from the original image to obtain vertical ridge lines of the plot to be detected; Based on the vertical ridge line and the boundary box, obtaining the seedling missing rate of the original image; The seedling absence rates of a plurality of the original images are used to obtain an average value of the seedling absence rates, and the average value is used as the seedling absence rate of the plot to be detected.

2. The method for calculating the seedling shortage rate according to claim 1, wherein: Before obtaining the sample images of the crops of the plot to be detected and the mapping relationship taken in advance, the method further includes: Acquire the original image and construct a rectangular coordinate system; wherein the origin of the rectangular coordinate system is the point at the upper left corner of the original image; Cut the original images in order without overlapping to obtain a plurality of sample images, and record the cut coordinates of the upper left corner of the sample image in the rectangular coordinate system; The mapping relationship is obtained by using the clipping coordinates.

3. The calculation method of the seedling shortage rate according to claim 2, characterized in that: The obtaining of the original image specifically includes: Obtain color images taken by the drone when it flies along the vertical ridge line; The edge cutting is performed based on the center point of the color image to obtain a plurality of the original images.

4. The method for calculating the seedling shortage rate according to claim 1, wherein: Extracting the ridge lines of the original image to obtain the vertical ridge lines of the plot to be detected specifically includes: The original image is sequentially subjected to graying, Gaussian filtering, edge detection, vertical line detection based on Hough transform and morphological operation to extract ridge lines, thereby obtaining the vertical ridge lines of the plot to be detected.

5. The method for calculating the seedling shortage rate according to claim 1, wherein: The step of obtaining the seedling absence rate of the original image by using the vertical ridge line and the boundary box includes: Calculating the total length of the vertical ridge lines and the sum of the heights of the boundary box; The ratio of the total height to the total length is taken as the missing seedling rate of the original image.

6. The method for calculating the seedling shortage rate according to claim 1, wherein: The training process of the seedling-deficient region detection model is as follows: Constructing a data set, the data set includes a labeled sample image and an unlabeled sample image, the label information of the labeled sample image includes a first bounding box and a category of the seedling-missing area, wherein the category is the seedling-missing area and the background area; Input the labeled sample image into the neural network model to obtain a first prediction box of the missing seedling area, calculate a first loss function according to the first prediction box and the first bounding box, update the parameters of the neural network model according to the first loss function, obtain a fully supervised target detection model, and use the fully supervised target detection model as a teacher model and a student model; Input the unlabeled sample image into the teacher model to obtain a sample image with a pseudo label, input the sample image with the pseudo label into the student model to obtain a second prediction box of the seedling-missing area; wherein the pseudo label includes a second bounding box marking the seedling-missing area; A second loss function is calculated according to the second prediction box and the second bounding box, and the parameters of the student model are updated according to the second loss function to obtain a missing seedling area detection model.

7. A system for calculating the seedling shortage rate, characterized in that: The calculation system of the seedling shortage rate comprises: An acquisition module, used to acquire a sample image of crops of a plot to be detected that has been taken in advance and a mapping relationship; wherein the mapping relationship is a positional relationship between the sample image and an original image; and the original image contains a plurality of sample images; A prediction module is used to detect the missing seedling area of ​​the sample image using a missing seedling area detection model to obtain a prediction frame of the missing seedling area; wherein the missing seedling area detection model is obtained by performing full supervision and semi-supervised training on a neural network model; A seedling-missing region determination module, used to obtain a boundary frame of the seedling-missing region in the original image by using the mapping relationship and the prediction frame; A ridge line extraction module is used to extract ridge lines from the original image to obtain vertical ridge lines of the plot to be detected; A first calculation module is used to obtain the seedling absence rate of the original image based on the vertical field ridge line and the boundary box; The second calculation module is used to use the seedling absence rates of multiple original images to obtain an average value of the seedling absence rates, and use the average value as the seedling absence rate of the plot to be detected.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for calculating the seedling deficiency rate according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for calculating the seedling deficiency rate according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method for calculating the seedling deficiency rate according to any one of claims 1 to 6 are implemented.

Citation Information

Patent Citations

  • High-resolution field image rice ear detection and counting method based on deep learning

    CN112069985A

  • Corn seedling missing rate calculation method based on deep learning of remote sensing image of unmanned aerial vehicle

    CN115631435A

  • Seedling missing identification method, device, equipment and medium

    CN117636314A

  • Target detection method and device, computer equipment and computer readable storage medium

    CN118262189A