Crop photosynthetic parameter prediction method and device based on deep learning, medium and product

Abstract

The invention discloses a crop photosynthetic parameter prediction method and device based on deep learning, a medium and a product, and relates to the field of photosynthetic parameter prediction, and the method comprises the steps: obtaining a hyperspectral data cube of a to-be-predicted crop; according to the hyperspectral data cube, determining the hyperspectral reflectivity of the to-be-predicted crop; according to the hyperspectral reflectivity, utilizing a photosynthetic parameter prediction model to determine photosynthetic parameters of the to-be-predicted crop; the photosynthetic parameters comprise a photosynthetic rate, a chlorophyll relative content, chlorophyll a, chlorophyll b and a leaf nitrogen content under specific light intensity; the photosynthetic parameter prediction model is obtained by training a deep learning model by using a training data set. According to the method, the photosynthetic parameter prediction model is determined by using the true values of the photosynthetic parameters and the corresponding hyperspectral reflectivity, so that the plurality of photosynthetic parameters of the crops can be determined quickly, losslessly and accurately.

Description

Technical Field

[0001] The present application relates to the field of photosynthetic parameter prediction, and in particular to a crop photosynthetic parameter prediction method, equipment, medium and product based on deep learning. Background Art

[0002] Photosynthetic rate under specific light intensity, SPAD (Soil and Plant Analyzer Development, relative chlorophyll content), chlorophyll a, chlorophyll b and leaf nitrogen content are key indicators for evaluating crop photosynthetic efficiency. Realizing rapid and non-destructive detection of these photosynthetic parameters is not only an important prerequisite for analyzing the physiological mechanism of high photosynthetic efficiency of crops, but also a core means to cultivate high photosynthetic efficiency crops and increase crop yields. Traditional methods for measuring photosynthetic parameters usually rely on manual operation and use a variety of equipment to measure separately. These methods are not only complicated to operate, time-consuming and labor-intensive, but some of the methods are destructive detection, which causes certain damage to the measured objects. In addition, these methods require high professional skills of operators, which limits their application in large-scale crop detection.

[0003] With the development of phenomics, hyperspectral imaging technology has gradually become an effective tool for detecting crop physiological traits. Hyperspectral imaging equipment uses sensors to obtain continuous narrow-band spectral data from visible light to near-infrared spectra, which can capture a large amount of spatial and spectral details and provide a non-destructive detection method for plant physiological traits. This technology enables some physiological parameters of plants that are invisible to the naked eye to be quickly obtained, greatly improving detection efficiency.

[0004] However, the spectral analysis method based on traditional machine learning is highly dependent on artificial feature engineering in the process of photosynthetic parameter prediction. Researchers usually need to conduct a large number of experiments to select appropriate spectral preprocessing methods and prediction models. This process is not only time-consuming and labor-intensive, but also has the risk of misuse of preprocessing, which may lead to a decrease in the prediction accuracy of the model. Summary of the invention

[0005] The purpose of this application is to provide a crop photosynthetic parameter prediction method, equipment, medium and product based on deep learning to improve the prediction accuracy of crop photosynthetic parameters.

[0006] To achieve the above objectives, this application provides the following solutions: In a first aspect, the present application provides a crop photosynthetic parameter prediction method based on deep learning, comprising: Obtaining a hyperspectral data cube of the crop to be predicted; the hyperspectral data cube includes a two-dimensional image of all bands; the value of each pixel point on the two-dimensional image is a digital intensity value of the hyperspectral sensor; Determining the hyperspectral reflectance of the crop to be predicted according to the hyperspectral data cube; According to the hyperspectral reflectance, the photosynthetic parameters of the crop to be predicted are determined using a photosynthetic parameter prediction model; the photosynthetic parameters include photosynthetic rate, relative chlorophyll content, chlorophyll a, chlorophyll b and leaf nitrogen content under a specific light intensity; wherein the photosynthetic parameter prediction model is obtained by training a deep learning model using a training data set; the training data set includes the hyperspectral reflectance of the training crop and the corresponding photosynthetic parameters; the deep learning model includes a Transformer module, an Inception module and a decoder; the Transformer module and the Inception module are both connected to the decoder.

[0007] Optionally, determining the hyperspectral reflectance of the crop to be predicted according to the hyperspectral data cube specifically includes: According to the hyperspectral data cube, the region of interest is identified using ENVI software; the region of interest is the region where the target leaf is located; the target leaf is the leaf for photosynthetic parameter prediction in the crop to be predicted; Determine the reflectivity of the corresponding pixel points based on the digital intensity values ​​of all pixel points in the region of interest; The hyperspectral reflectance of the crop to be predicted is determined based on the reflectance of all the pixels in the region of interest; the hyperspectral reflectance is the average value of the reflectance of all the pixels in the region of interest.

[0008] Optionally, determining the reflectivity of corresponding pixels based on the digital intensity values ​​of all pixels in the region of interest specifically includes: Using the formula Determine the reflectivity of the pixel; in, is the spatial position of the pth pixel and wavelength The reflectivity at is the spatial position of the pth pixel and wavelength The digital intensity value at ; The black reference plate is located in space and wavelength The digital intensity value at ; The white reference plate is located in space and wavelength The digital intensity value at ; The white reference plate is located in space and wavelength The reflectivity at .

[0009] Optionally, the Transformer module includes a self-attention layer, a feedforward network layer and a normalization layer connected in sequence; the feedforward network layer includes two fully connected layers.

[0010] Optionally, the Inception module includes a first branch, a second branch, a third branch and a fourth branch; The first branch includes 3 convolutional layers; the second branch includes 2 convolutional layers; the third branch includes 1 maximum pooling layer and 1 convolutional layer; the fourth branch includes 1 convolutional layer.

[0011] Optionally, the decoder comprises three fully connected layers connected in sequence.

[0012] Optionally, the deep learning model is trained using the training data set to obtain a photosynthetic parameter prediction model, specifically including: Get the training dataset; Inputting the hyperspectral reflectance of the training crop into the current deep learning model to obtain predicted values ​​of photosynthetic parameters of the training crop; Determining a loss function value using a multi-task loss function according to the predicted value of the photosynthetic parameter of the training crop and the photosynthetic parameter corresponding to the hyperspectral reflectance of the training crop; Determine whether a training end condition is met; the training end condition is that the loss function value is less than a set value; If yes, the current deep learning model is used as the photosynthetic parameter prediction model; If not, the network parameters of the current deep learning model are adjusted based on the loss function value, and the method returns "inputting the hyperspectral reflectance of the training crop into the current deep learning model to obtain the predicted value of the photosynthetic parameter of the training crop".

[0013] In a second 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 any of the above-mentioned methods for predicting crop photosynthetic parameters based on deep learning.

[0014] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned methods for predicting crop photosynthetic parameters based on deep learning.

[0015] In a fourth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned deep learning-based crop photosynthetic parameter prediction methods.

[0016] According to the specific embodiments provided in this application, this application has the following technical effects: The present application provides a crop photosynthetic parameter prediction method, device, medium and product based on deep learning, which obtains a hyperspectral data cube of the crop to be predicted; determines the hyperspectral reflectance of the crop to be predicted based on the hyperspectral data cube; determines the photosynthetic parameters of the crop to be predicted using a photosynthetic parameter prediction model based on the hyperspectral reflectance; the photosynthetic parameters include photosynthetic rate, relative chlorophyll content, chlorophyll a, chlorophyll b and leaf nitrogen content under a specific light intensity; the photosynthetic parameter prediction model is obtained by training a deep learning model using a training data set. In the present application, the true value of the photosynthetic parameter and the corresponding hyperspectral reflectance are used to determine the photosynthetic parameter prediction model, which can quickly, non-destructively and accurately determine multiple photosynthetic parameters of the crop. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] 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.

[0018] Figure 1 A schematic diagram of a process flow of a crop photosynthetic parameter prediction method based on deep learning provided in one embodiment of the present application; Figure 2 A schematic diagram of the structure of the deep learning model provided for this application; Figure 3 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0019] 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.

[0020] 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 with reference to the accompanying drawings and specific implementation methods.

[0021] In an exemplary embodiment, Figure 1 As shown, a crop photosynthetic parameter prediction method based on deep learning is provided, comprising the following steps: S1: Obtain a hyperspectral data cube of the crop to be predicted; the hyperspectral data cube includes a two-dimensional image of all bands; the value of each pixel point on the two-dimensional image is a digital intensity value of a hyperspectral sensor.

[0022] In practical applications, hyperspectral sensors are used to obtain hyperspectral data cubes of crops to be predicted. The hyperspectral data cube consists of continuous bands, each band is a two-dimensional image, and the value of each pixel on the two-dimensional image is the digital intensity value of the hyperspectral sensor.

[0023] S2: Determine the hyperspectral reflectance of the crop to be predicted according to the hyperspectral data cube.

[0024] As an optional implementation, S2 specifically includes: S21: According to the hyperspectral data cube, using ENVI software, identifying a region of interest; the region of interest is the region where the target leaves are located; the target leaves are the leaves for which photosynthetic parameters are predicted in the crop to be predicted.

[0025] In practical applications, the ENVI software is used to manually identify the region of interest for the hyperspectral data cube, where the region of interest is the target leaf for which the photosynthetic parameters are to be predicted.

[0026] S22: Determine the reflectivity of corresponding pixels based on the digital intensity values ​​of all pixels in the region of interest.

[0027] As an optional implementation, S22 specifically includes: Using the formula Determines the reflectivity of a pixel.

[0028] in, is the spatial position of the pth pixel and wavelength The reflectivity at is the spatial position of the pth pixel and wavelength The digital intensity value at ; The black reference plate is located in space and wavelength The digital intensity value at ; The white reference plate is located in space and wavelength The digital intensity value at ; The white reference plate is located in space and wavelength The reflectivity at . The white reference plate is made of a material with good Lambertian properties.

[0029] When the reflectivity of the white reference plate is higher than 95%, the above reflectivity calculation formula can be written as: .

[0030] S23: Determine the hyperspectral reflectance of the crop to be predicted based on the reflectance of all the pixels in the region of interest; the hyperspectral reflectance is the average value of the reflectance of all the pixels in the region of interest.

[0031] S3: According to the hyperspectral reflectance, using a photosynthetic parameter prediction model, determine the photosynthetic parameters of the crop to be predicted; the photosynthetic parameters include photosynthetic rate, relative chlorophyll content (Soil and Plant Analyzer Development, SPAD), chlorophyll a, chlorophyll b and leaf nitrogen content under a specific light intensity; wherein, the photosynthetic parameter prediction model is obtained by training a deep learning model using a training data set; the training data set includes the hyperspectral reflectance of the training crop and the corresponding photosynthetic parameters; the deep learning model includes a Transformer module, an Inception module and a decoder; the Transformer module and the Inception module are both connected to the decoder.

[0032] The model structure of the deep learning model is as follows Figure 2 As shown, the Transformer module includes a self-attention layer, a feedforward network layer and a normalization layer connected in sequence; the feedforward network layer includes two fully connected layers. The self-attention layer, the feedforward network layer and the normalization layer constitute a unit, which is repeated three times to form a Transformer module.

[0033] The Inception module includes a first branch (branch 1), a second branch (branch 2), a third branch (branch 3) and a fourth branch (branch 4).

[0034] Branch 1 contains 3 convolutional layers, branch 2 contains 2 convolutional layers, branch 3 contains 1 maximum pooling layer and 1 convolutional layer, and branch 4 contains 1 convolutional layer. Each convolutional layer uses the RELU activation function.

[0035] The decoder includes three fully connected layers connected in sequence. The first fully connected layer and the second fully connected layer use the RELU activation function, and the third fully connected layer does not use an activation function.

[0036] As an optional implementation method, the deep learning model is trained using the training data set to obtain a photosynthetic parameter prediction model, which specifically includes: Get the training dataset.

[0037] The hyperspectral reflectance of the training crop is input into the current deep learning model to obtain the predicted values ​​of the photosynthetic parameters of the training crop.

[0038] The loss function value is determined by using a multi-task loss function according to the predicted value of the photosynthetic parameter of the training crop and the photosynthetic parameter corresponding to the hyperspectral reflectance of the training crop.

[0039] Determine whether a training end condition is met; the training end condition is that the loss function value is less than a set value.

[0040] If so, the current deep learning model is used as the photosynthetic parameter prediction model.

[0041] If not, the network parameters of the current deep learning model are adjusted based on the loss function value, and the method returns "inputting the hyperspectral reflectance of the training crop into the current deep learning model to obtain the predicted value of the photosynthetic parameter of the training crop".

[0042] In practical applications, the specific training process of the deep learning model is as follows: Obtain hyperspectral reflectance and the corresponding true values ​​of photosynthetic parameters.

[0043] The hyperspectral reflectance is input into the current deep learning model, and the training optimization strategy is trained for specific rounds until the training loss remains basically stable at a small value.

[0044] The above training optimization strategy refers to multi-task loss optimization of the model. The calculation formula of the loss function is: .

[0045] in, For the The loss of a task, For the The weight of the task, To account for the overall loss of the model, the weight of each task is adjusted using the GradNorm algorithm.

[0046] Among them, the specific training process of the Transformer module includes: The hyperspectral reflectance is input into the attention layer and first transformed through three linear layers to generate query, key and value. The attention weight is obtained by applying the Softmax function to the dot product of the query and the key, and then the weight is multiplied by the value to get the weighted sum.

[0047] The obtained weighted sum is added to the original spectral reflectance to realize the first residual connection, and the result of the first residual connection is normalized.

[0048] The normalized result is input into the feedforward network layer to obtain the output of the feedforward network layer. The feedforward network layer is composed of two fully connected layers. The first fully connected layer uses the GELU activation function, and the second fully connected layer does not use the activation function. Each fully connected layer output is randomly inactivated.

[0049] The output of the feedforward network layer is added to the result of the first residual connection to realize the second residual connection, and the result of the second residual connection is normalized.

[0050] The specific training process of the Inception module includes: The hyperspectral reflectance is input into four branches respectively to obtain four outputs, and the four outputs are concatenated to obtain the output of the Inception module.

[0051] Specifically, the hyperspectral reflectance is input into branch 1 of the Inception module, including: The hyperspectral reflectance is input to the first convolutional layer.

[0052] The output of the first convolutional layer uses the RELU activation function.

[0053] The result of using the RELU activation function in the previous step is input into the second convolutional layer.

[0054] The output of the second convolutional layer uses the RELU activation function.

[0055] The result of using the RELU activation function in the previous step is input into the third convolutional layer.

[0056] The output of the third convolutional layer uses the RELU activation function.

[0057] The hyperspectral reflectance is input to branch 2 of the Inception module, including: The hyperspectral reflectance is input to the first convolutional layer.

[0058] The output of the first convolutional layer uses the RELU activation function.

[0059] The result of using the RELU activation function in the previous step is input into the second convolutional layer.

[0060] The output of the second convolutional layer uses the RELU activation function.

[0061] The hyperspectral reflectance is input to branch 3 of the Inception module, including: The hyperspectral reflectance is input into the max pooling layer.

[0062] The output of the max pooling layer is input into the convolutional layer.

[0063] The output of the convolutional layer uses the RELU activation function.

[0064] The hyperspectral reflectance is input to branch 4 of the Inception module, including: The hyperspectral reflectance is input to the convolutional layer.

[0065] The output of the convolutional layer uses the RELU activation function.

[0066] The output results of the four branches are feature concatenated.

[0067] The specific training process of the Decoder includes: The global features extracted by the Transformer module and the local features extracted by the Inception module are added and fused.

[0068] The fused features are input into the Decoder to obtain the predicted values ​​of multiple photosynthetic parameters.

[0069] The features obtained by the Transformer module and the Inception module are fused and input into the Decoder decoder, including: The fused features are input into the first fully connected layer.

[0070] The output of the first fully connected layer uses the RELU activation function.

[0071] The result of using the RELU activation function in the previous step is input into the second fully connected layer.

[0072] The output of the second fully connected layer uses the RELU activation function.

[0073] The result of using the RELU activation function in the previous step is input into the third fully connected layer.

[0074] The output of the third fully connected layer uses the RELU activation function.

[0075] This application first obtains the hyperspectral data cube of the crop to be predicted; then, manually identifies the range of the crop to be predicted through ENVI software, and obtains the hyperspectral reflectance of the crop to be predicted based on the reflectance calculation formula; finally, the hyperspectral reflectance value of the crop to be predicted is input into the photosynthetic parameter prediction model based on deep learning to obtain the predicted value of the corresponding photosynthetic parameter; photosynthetic parameters include: photosynthetic rate, SPAD, chlorophyll a, chlorophyll b and leaf nitrogen content under specific light intensity. In this application, the true value of the photosynthetic parameter and the corresponding hyperspectral reflectance are used to determine the photosynthetic parameter prediction model based on deep learning, which can quickly, non-destructively and accurately determine multiple photosynthetic parameters of crops.

[0076] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above-mentioned crop photosynthetic parameter prediction method based on deep learning when executing the computer program.

[0077] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the above-mentioned crop photosynthetic parameter prediction method based on deep learning.

[0078] In an exemplary embodiment, a computer program product is provided, including a computer program, which, when executed by a processor, implements the above-mentioned crop photosynthetic parameter prediction method based on deep learning.

[0079] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a crop photosynthetic parameter prediction method based on deep learning is implemented.

[0080] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0081] 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.

[0082] Those skilled 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 this 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), magnetoresistive 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).

[0083] 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.

[0084] 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.

[0085] 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 crop photosynthetic parameter prediction method based on deep learning, characterized in that: include: Obtain the hyperspectral data cube of the crop to be predicted; The hyperspectral data cube includes a two-dimensional image of all bands; the value of each pixel on the two-dimensional image is a digital intensity value of the hyperspectral sensor; Determining the hyperspectral reflectance of the crop to be predicted according to the hyperspectral data cube; According to the hyperspectral reflectance, the photosynthetic parameters of the crop to be predicted are determined using a photosynthetic parameter prediction model; the photosynthetic parameters include photosynthetic rate, relative chlorophyll content, chlorophyll a, chlorophyll b and leaf nitrogen content under a specific light intensity; wherein the photosynthetic parameter prediction model is obtained by training a deep learning model using a training data set; the training data set includes the hyperspectral reflectance of the training crop and the corresponding photosynthetic parameters; the deep learning model includes a Transformer module, an Inception module and a decoder; the Transformer module and the Inception module are both connected to the decoder.

2. The crop photosynthetic parameter prediction method based on deep learning according to claim 1, characterized in that: Determining the hyperspectral reflectance of the crop to be predicted according to the hyperspectral data cube specifically includes: According to the hyperspectral data cube, the region of interest is identified using ENVI software; the region of interest is the region where the target leaf is located; the target leaf is the leaf for photosynthetic parameter prediction in the crop to be predicted; Determine the reflectivity of the corresponding pixel points based on the digital intensity values ​​of all pixel points in the region of interest; The hyperspectral reflectance of the crop to be predicted is determined based on the reflectance of all the pixels in the region of interest; the hyperspectral reflectance is the average value of the reflectance of all the pixels in the region of interest.

3. The crop photosynthetic parameter prediction method based on deep learning according to claim 2, characterized in that: Determining the reflectivity of corresponding pixel points based on the digital intensity values ​​of all pixel points in the region of interest specifically includes: Using the formula Determine the reflectivity of the pixel; in, is the spatial position of the pth pixel and wavelength The reflectivity at is the spatial position of the pth pixel and wavelength The digital intensity value at ; The black reference plate is located in space and wavelength The digital intensity value at ; The white reference plate is located in space and wavelength The digital intensity value at ; The white reference plate is located in space and wavelength The reflectivity at .

4. The crop photosynthetic parameter prediction method based on deep learning according to claim 1, characterized in that: The Transformer module includes a self-attention layer, a feedforward network layer and a normalization layer connected in sequence; the feedforward network layer includes two fully connected layers.

5. The crop photosynthetic parameter prediction method based on deep learning according to claim 1, characterized in that: The Inception module includes a first branch, a second branch, a third branch and a fourth branch; The first branch includes 3 convolutional layers; the second branch includes 2 convolutional layers; the third branch includes 1 maximum pooling layer and 1 convolutional layer; the fourth branch includes 1 convolutional layer.

6. The crop photosynthetic parameter prediction method based on deep learning according to claim 1, characterized in that: The decoder includes 3 fully connected layers connected in sequence.

7. The crop photosynthetic parameter prediction method based on deep learning according to claim 1, characterized in that: The deep learning model is trained using the training data set to obtain a photosynthetic parameter prediction model, which includes: Get the training dataset; Inputting the hyperspectral reflectance of the training crop into the current deep learning model to obtain predicted values ​​of photosynthetic parameters of the training crop; Determining a loss function value using a multi-task loss function according to the predicted value of the photosynthetic parameter of the training crop and the photosynthetic parameter corresponding to the hyperspectral reflectance of the training crop; Determine whether a training end condition is met; the training end condition is that the loss function value is less than a set value; If yes, the current deep learning model is used as the photosynthetic parameter prediction model; If not, the network parameters of the current deep learning model are adjusted based on the loss function value, and "the hyperspectral reflectance of the training crop is input into the current deep learning model to obtain the predicted value of the photosynthetic parameter of the training crop" is returned.

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 crop photosynthetic parameter prediction method based on deep learning as described in any one of claims 1 to 7.

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 crop photosynthetic parameter prediction method based on deep learning described in any one of claims 1 to 7 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the crop photosynthetic parameter prediction method based on deep learning described in any one of claims 1 to 7 is implemented.

Images