Crop seed vitality prediction method and system based on multi-source information coupling

Abstract

The present invention discloses a crop seed vitality prediction method and system based on multi-source information coupling, including: using a spectral camera to obtain spectral data of seeds, extracting appearance phenotypic data, biochemical parameters, environmental data and real-time dynamic 3D phenotypic data of seeds, and generating multi-source data of seeds; dividing the multi-source data into static data and dynamic data, embedding and encoding the input data in a seed vitality prediction model, introducing dual-channel attention to fuse and learn the static data vector and the dynamic data vector, and obtaining a fusion vector; obtaining the prediction results of seed germination rate and seedling rate through a fully connected layer, and comprehensively analyzing and obtaining seed vitality. The present invention integrates a seed vitality prediction model of multi-source information fusion and dual-channel attention mechanism into a seed vitality prediction system, realizes efficient and accurate prediction of seed vitality, and provides a scientific basis for agricultural production.

Description

Technical Field

[0001] The present invention relates to the technical field of big data intelligent analysis and seed quality detection, and more specifically, to a crop seed vitality prediction method and system based on multi-source information coupling. Background Art

[0002] Rapid and accurate detection of seed quality is a key link in ensuring seed quality. It can ensure smooth and orderly food production from the source, reduce agricultural production risks, and promote agricultural scientific and technological progress. Seed vigor, as an important indicator of seed quality, directly affects the germination rate, growth rate and final yield of crops. The assessment of seed vigor mainly relies on germination experiments under laboratory conditions. This method is time-consuming and costly, and it is difficult to fully reflect the performance of seeds in complex natural environments. Therefore, it is particularly important to develop an efficient, accurate and fast seed vigor prediction technology.

[0003] The rapid development of the computer field has provided powerful data processing capabilities for the study of high-throughput seeds. At present, more and more artificial intelligence technologies are being applied to the field of seed research, such as using hyperspectral imaging technology to evaluate the vitality of seeds by analyzing their spectral characteristics, and using infrared spectroscopy technology to predict their germination ability by measuring the chemical composition of seeds. However, these methods are not flexible and do not take into account the fact that seed vitality will change over time. Therefore, how to integrate seed quality phenotypic parameter data, environmental data, genetic data, and real-time dynamic 3D phenotypic data of different germination degrees, construct a dynamic prediction model for seed vitality, and accurately predict seed vitality and analyze the laws of seed vitality are core technical issues that need to be solved urgently. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a crop seed vitality prediction method and system based on multi-source information coupling, the purpose of which is to predict the plant germination rate and seedling rate of seeds through seed phenotypic data and spectral data, and analyze seed vitality.

[0005] A first aspect of the present invention provides a crop seed vitality prediction method based on multi-source information coupling, comprising:

[0006] Using a spectral camera to obtain spectral data of seeds, extracting appearance phenotypic data, biochemical parameters, environmental data and real-time dynamic 3D phenotypic data of seeds based on the spectral data, and generating multi-source data of seeds;

[0007] Dividing the multi-source data into static data and dynamic data, constructing a seed vitality prediction model, and using the static data and the dynamic data as model input;

[0008] In the seed vitality prediction model, input data is embedded and encoded to generate a static data vector and a dynamic data vector, and dual-channel attention is introduced to perform fusion learning on the static data vector and the dynamic data vector to obtain a fusion vector;

[0009] The fusion vector is used to obtain the prediction results of seed germination rate and seedling rate through a fully connected layer, and the seed vitality is represented by the prediction results of seed germination rate and seedling rate.

[0010] In this solution, the appearance phenotypic data, biochemical parameters, environmental data and real-time dynamic 3D phenotypic data of the seeds are extracted based on the spectral data to generate multi-source data of the seeds, specifically:

[0011] The collected spectral data and phenotypic data are preprocessed, and the SAM large model is called to perform fine-grained instance segmentation on the seeds to obtain the seed appearance phenotypic data;

[0012] Based on the graph neural network and LSTM network component space-time graph network, the space-time graph network model is used to extract data on the biochemical parameters of seeds;

[0013] Using a multi-view diffusion model and a sparse view large-scale reconstruction model to obtain seed three-dimensional grid data, and combining the three-dimensional grid data with a seed germination growth example to obtain environmental data during seed germination and real-time dynamic 3D phenotypic data of different germination degrees;

[0014] Aggregate appearance phenotypic data, biochemical parameters, and environmental data sets to generate multi-source data of seeds in real-time dynamic 3D phenotypic data.

[0015] In this scheme, a seed vitality prediction model is constructed, and the seed vitality prediction model includes a feature encoding module, a dual-channel attention aggregation module and a prediction module;

[0016] The feature encoding module is used for static data and dynamic data Perform embedded coding to obtain static data vector and dynamic data vector .

[0017] In this solution, the dual-channel attention aggregation module performs fusion learning on static data vectors and dynamic data vectors, specifically:

[0018] The static data vector and dynamic data vector Perform position encoding to obtain the static data vector after position encoding and dynamic data vector ;

[0019] The position-encoded static data vector and dynamic data vector Perform different partial mask operations separately to obtain the mask vector and mask vector ;

[0020] The mask vector Import it into Transformer, use the masked self-attention mechanism to get the output of the single-head attention, concatenate the outputs of all attention heads, and obtain the output of the masked multi-head attention through linear transformation;

[0021] The output of the mask multi-head attention is compared with the corresponding mask vector Add and normalize to get and ;

[0022] Will obtain and Input into the feedforward neural network to obtain the output results, and the respective output results are and After addition, normalization is performed to obtain the vector and ,Will and Add together to get the fusion vector .

[0023] In this scheme, in the masked self-attention mechanism in Transformer, the calculation of single-head attention is expressed as:

[0024] ;

[0025] in, Indicates Attention head, represent query vector, key vector and value vector respectively, They represent learnable parameters respectively;

[0026] The outputs of all attention heads are concatenated and the masked multi-head attention output is obtained through linear transformation, which is expressed as:

[0027] ;

[0028] in represents a learnable parameter;

[0029] The output of the masked multi-head attention is added to the respective embedding vectors and normalized. The vector It is expressed as:

[0030] ;

[0031] in represents the output of masked multi-head attention, represents the original embedding matrix.

[0032] In this scheme, the fusion vector After calculation by the fully connected layer, the prediction results of germination rate and seedling rate are obtained, and the prediction results of germination rate and seedling rate are adaptively weighted. The prediction results of seed vigor are output by weighted summation of the germination rate prediction results and the seedling rate prediction results.

[0033] The second aspect of the present invention provides a crop seed vitality prediction system based on multi-source information coupling, comprising: a data acquisition unit, a model aggregation unit and an information interaction unit;

[0034] The data acquisition unit collects multi-source heterogeneous spectral data through hardware equipment, aggregates the multi-source heterogeneous spectral data and parses out seed phenotypic data;

[0035] The model aggregation unit includes multiple deep learning models, including a SAM large model, a spatiotemporal graph network model, a multi-view diffusion model, a sparse view large-scale reconstruction model, and a seed vitality prediction model constructed by a dual-channel attention aggregation neural network, which extracts the appearance phenotypic data, biochemical parameters, environmental data, and real-time dynamic 3D phenotypic data of seeds based on spectral data and phenotypic data to generate multi-source data of seeds;

[0036] The information interaction unit includes a background management module and a front-end interaction module. The background management module is responsible for user role information management, log management, and data management; the front-end interaction module is responsible for data visualization display.

[0037] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0038] The present invention constructs dual-channel data, namely static data and dynamic data. A horizontal dot multiplication operation is performed on the static data vector to capture the relationship between features; and the dynamic data is analyzed using cross-correlation. The combination of dynamic data and static data allows the model to learn richer feature information and better analyze and predict data.

[0039] The present invention constructs a crop seed vitality prediction model that integrates multi-source information. A transformer encoder is used to capture the global information of static data and the time dependency of dynamic data. The characteristics of the two types of data are combined to improve the prediction ability of the model, thereby achieving efficient and accurate prediction of plant germination rate and seedling rate.

[0040] The present invention designs a system for predicting crop seed vitality, which realizes the automatic collection and analysis of relevant data such as seed color texture and structural morphology, provides a data basis for seed vitality prediction, and realizes intelligent seed vitality prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments or exemplary embodiments of the present invention, the drawings required for use in the embodiments or exemplary descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained according to the drawings without paying creative work.

[0042] Figure 1 A flow chart of a crop seed vitality prediction method based on multi-source information coupling is shown;

[0043] Figure 2 A schematic diagram of a seed vigor prediction model is shown;

[0044] Figure 3 The framework diagram of the crop seed vitality prediction system based on multi-source information coupling is shown;

[0045] Figure 4 A schematic diagram of an intelligent crop seed monitoring device is shown. DETAILED DESCRIPTION

[0046] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0048] Figure 1 A flow chart of a crop seed vitality prediction method based on multi-source information coupling is shown.

[0049] like Figure 1 As shown, this embodiment provides a crop seed vitality prediction method based on multi-source information coupling, including:

[0050] S102, using a spectral camera to obtain spectral data of seeds, extracting appearance phenotypic data, biochemical parameters, environmental data and real-time dynamic 3D phenotypic data of seeds based on the spectral data, and generating multi-source data of seeds;

[0051] S104, dividing the multi-source data into static data and dynamic data, constructing a seed vitality prediction model, and using the static data and the dynamic data as model input;

[0052] S106, embedding and encoding the input data in the seed vitality prediction model to generate a static data vector and a dynamic data vector, introducing dual-channel attention to perform fusion learning on the static data vector and the dynamic data vector, and obtaining a fusion vector;

[0053] S108, using the fusion vector to obtain seed germination rate and seedling rate prediction results through a fully connected layer, and using the germination rate and seedling rate prediction results to characterize seed vitality.

[0054] It should be noted that the collected spectral data and phenotypic data are preprocessed, and the SAM large model is called to perform fine-grained instance segmentation on the seeds to obtain the appearance phenotypic data of the seeds such as the hilum, epidermal cracks, damage, and diseases; a spatiotemporal graph network is established based on the graph neural network and the LSTM network, and the spatiotemporal graph network model is used to extract data on biochemical parameters such as seed water content, protein content, starch content, and oil content; a multi-view diffusion model and a sparse view large-scale reconstruction model are used to obtain seed three-dimensional grid data, and the environmental data in the seed germination process and real-time dynamic 3D phenotypic data of different germination degrees are obtained in combination with the three-dimensional grid data according to the seed germination growth example; the appearance phenotypic data, biochemical parameters, and real-time dynamic 3D phenotypic data of the environmental data set are aggregated to generate multi-source data of seeds.

[0055] Figure 2 A schematic diagram of the seed vigor prediction model is shown.

[0056] Construct a seed vitality prediction model, which includes a feature encoding module, a dual-channel attention aggregation module and a prediction module; the feature encoding module performs static data and dynamic data Perform embedded coding to obtain static data vector and dynamic data vector Static data include seed appearance phenotypic data and biochemical parameters, and dynamic data include seed environmental data and real-time dynamic 3D phenotypic data.

[0057] The dual-channel attention aggregation module performs fusion learning on the divided dynamic data and static data to capture the complex interactions and dependencies between them. and dynamic data vector Perform position encoding to obtain the static data vector after position encoding and dynamic data vector , the position coding calculation formula is:

[0058] ;

[0059] in denote the values ​​of odd and even positions in the position encoding matrix, respectively, represents the position of the input sequence, represents the dimension of the position encoding, Represents the dimension of the input sequence.

[0060] The position-encoded static data vector and dynamic data vector Perform different partial mask operations separately to obtain the mask vector and mask vector ;

[0061] The mask vector Import it into Transformer and use the masked self-attention mechanism to get the output of single-head attention. The calculation of single-head attention is expressed as:

[0062] ;

[0063] in, Indicates Attention head, represent query vector, key vector and value vector respectively, They represent learnable parameters respectively;

[0064] The outputs of all attention heads are concatenated, and the output of the masked multi-head attention is obtained through linear transformation, which is expressed as:

[0065] ;

[0066] in represents a learnable parameter;

[0067] The output of the mask multi-head attention is compared with the corresponding mask vector Add and normalize to get and vector It is expressed as:

[0068] ;

[0069] in represents the output of masked multi-head attention, represents the original embedding matrix.

[0070] Will obtain and Input into the feedforward neural network to obtain the output results, and the respective output results are and After addition, normalization is performed to obtain the vector and Will and Add together to get the fusion vector .

[0071] The fusion vector After calculation by the fully connected layer, the prediction results of germination rate and seedling rate are obtained, and the prediction results of germination rate and seedling rate are adaptively weighted. The prediction results of seed vigor are output by weighted summation of the germination rate prediction results and the seedling rate prediction results.

[0072] Figure 3 A framework diagram of a crop seed vitality prediction system based on multi-source information coupling is shown.

[0073] The second aspect of the present invention provides a crop seed vitality prediction system based on multi-source information coupling, comprising: a data acquisition unit, a model aggregation unit and an information interaction unit;

[0074] The data acquisition unit collects multi-source heterogeneous spectral data through hardware equipment, aggregates the multi-source heterogeneous spectral data and parses out seed phenotypic data;

[0075] The core task of the data acquisition unit is to collect data and integrate multi-source, heterogeneous spectral data collected from various hardware devices, and extract key seed phenotype data through a sophisticated analysis process. In the data layer, data is collected through the control of each hardware device by the main control board, and then the data is stored in the memory for easy call by the model layer. A flexible storage solution is used here to ensure that data can be quickly accessed and updated in real time.

[0076] The model aggregation unit includes multiple deep learning models, including a SAM large model, a spatiotemporal graph network model, a multi-view diffusion model, a sparse view large-scale reconstruction model, and a seed vitality prediction model constructed by a dual-channel attention aggregation neural network, which extracts the appearance phenotypic data, biochemical parameters, environmental data, and real-time dynamic 3D phenotypic data of seeds based on spectral data and phenotypic data to generate multi-source data of seeds;

[0077] The model aggregation unit includes the SAM large model, the spatiotemporal graph network model, the multi-view diffusion model, the sparse view model, and the dual-channel attention aggregation neural network model. SAM performs fine-grained instance segmentation on seeds to extract the appearance phenotypic features of seeds such as the hilum, epidermal cracks, damage, and diseases. The spatiotemporal graph network model extracts data on biochemical parameters such as seed water content, protein content, starch content, and oil content. The multi-view diffusion model and the sparse view large model reconstruction model generate seed three-dimensional grid data to obtain seed environmental data and real-time dynamic 3D phenotypic data. After data extraction by these models, the extracted seed quality phenotypic parameter data, environmental data and other multi-source information are fused to obtain static data and dynamic data. Static data includes seed phenotypic parameters, and dynamic data includes seed environmental data and real-time dynamic 3D phenotypic data. Subsequently, the data of these two channels are sent to the dual-channel attention aggregation neural network model to obtain the output of seed vitality prediction and return it to the application layer.

[0078] The information interaction unit includes a background management module and a front-end interaction module. The background management module is responsible for user role information management, log management, and data management; the front-end interaction module is responsible for data visualization display.

[0079] The information interaction unit serves as the user interaction interface of the system. The application layer includes two modules: the background management module and the front-end interaction interface. The background management module has functions such as user role information management, log management, and data management to ensure the stable operation of the system and data security. At the same time, the background management module is also responsible for integrating the model with the data to provide efficient and convenient service support for the front end. The front end focuses on data visualization, presenting analysis results to users in an intuitive and easy-to-understand way to enhance user experience. Users can control the operation of the system by clicking on the front-end interaction interface. The front end collects seed data and predicts seed vitality by calling the data collection unit and the model aggregation unit, and then returns the results of the model aggregation unit to the front-end interface for display.

[0080] Figure 4 A schematic diagram of an intelligent crop seed monitoring device is shown.

[0081] This embodiment provides an intelligent crop seed monitoring device, including: a main control board, a memory, a feeder, a spectral camera, a halogen lamp, a display, a transformer and an external power supply;

[0082] The feed inlet is used for placing and taking out seeds, and the seeds are placed in by pressing the feed inlet;

[0083] The spectral camera is used to obtain imaging information of seeds in different wavelength bands, and after capturing the imaging information, transmit it back to the memory for storage.

[0084] The halogen lamps are evenly distributed around the hyperspectral camera to illuminate the crop seeds and provide an internal lighting environment for the detection device;

[0085] The display is used to display the multi-dimensional data of crop seeds and the prediction results of seed vitality.

[0086] The feed port is used for users to put in and take out test seeds. When users need to test seed data, they only need to gently press the feed port to put in the material. In addition, there are 24 halogen lamps evenly distributed around the spectral camera. This design can make the internal light source evenly distributed, which is conducive to data collection. The user puts the crop seeds in the feed port and then starts the equipment. The main control board not only has precise control capabilities, but also can quickly respond to user operation instructions and control the halogen lamp to reach the preset optimal brightness conditions in a short time. Through the precise control of the main control board, the evenly distributed halogen lamp array is activated to create an ideal lighting environment for the hyperspectral camera. When the internal brightness of the device meets the preset conditions, the hyperspectral camera automatically starts and uses its highly sensitive sensor and advanced multi-band imaging technology to capture the hyperspectral information of crop seeds. After capturing the data, the hyperspectral camera transmits it back to the memory for storage. The memory uses high-speed reading and writing technology to ensure that the data can be recorded and saved quickly and accurately for subsequent analysis and processing. At the same time, users can view the detailed results of seed multidimensional data and vitality prediction data on the display.

[0087] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0088] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0089] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0090] Those skilled in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0091] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0092] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A crop seed vitality prediction method based on multi-source information coupling, characterized in that: The following steps are involved: Using a spectral camera to obtain spectral data of seeds, extracting appearance phenotypic data, biochemical parameters, environmental data and real-time dynamic 3D phenotypic data of seeds based on the spectral data, and generating multi-source data of seeds; Dividing the multi-source data into static data and dynamic data, constructing a seed vitality prediction model, and using the static data and the dynamic data as model input; In the seed vitality prediction model, input data is embedded and encoded to generate a static data vector and a dynamic data vector, and dual-channel attention is introduced to perform fusion learning on the static data vector and the dynamic data vector to obtain a fusion vector; The fusion vector is used to obtain the prediction results of seed germination rate and seedling rate through a fully connected layer, and the germination rate and seedling rate prediction results are used to represent the seed vitality; Constructing a seed vitality prediction model, wherein the seed vitality prediction model includes a feature encoding module, a dual-channel attention aggregation module and a prediction module; The feature encoding module is used for static data and dynamic data Perform embedded coding to obtain static data vector and dynamic data vector ; The dual-channel attention aggregation module performs fusion learning on the static data vector and the dynamic data vector, specifically: The static data vector and dynamic data vector Perform position encoding to obtain the static data vector after position encoding and dynamic data vector ; The position-encoded static data vector and dynamic data vector Perform different partial mask operations separately to obtain the mask vector and mask vector ; The mask vector , Import it into Transformer, use the masked self-attention mechanism to get the output of the single-head attention, concatenate the outputs of all attention heads, and obtain the output of the masked multi-head attention through linear transformation; The output of the mask multi-head attention is compared with the corresponding mask vector , Add and normalize to get and ; Will obtain and Input into the feedforward neural network to obtain the output results, and the respective output results are and After addition, normalization is performed to obtain the vector and ,Will and Add together to get the fusion vector ; The fusion vector After calculation by the fully connected layer, the prediction results of germination rate and seedling rate are obtained, and the prediction results of germination rate and seedling rate are adaptively weighted. The prediction results of seed vigor are output by weighted summation of the germination rate prediction results and the seedling rate prediction results.

2. A crop seed vitality prediction method based on multi-source information coupling according to claim 1, characterized in that: Based on the spectral data, the appearance phenotypic data, biochemical parameters, environmental data and real-time dynamic 3D phenotypic data of the seeds are extracted to generate multi-source data of the seeds, specifically: The collected spectral data and phenotypic data are preprocessed, and the SAM large model is called to perform fine-grained instance segmentation on the seeds to obtain the seed appearance phenotypic data; A spatiotemporal graph network was constructed based on graph neural network and LSTM network, and the spatiotemporal graph network model was used to extract data on the biochemical parameters of seeds; Using a multi-view diffusion model and a sparse view large-scale reconstruction model to obtain seed three-dimensional grid data, and combining the three-dimensional grid data with a seed germination growth example to obtain environmental data during seed germination and real-time dynamic 3D phenotypic data of different germination degrees; Aggregate appearance phenotypic data, biochemical parameters, environmental data and real-time dynamic 3D phenotypic data to generate multi-source data for seeds.

3. A crop seed vitality prediction method based on multi-source information coupling according to claim 1, characterized in that: In the masked self-attention mechanism in Transformer, the calculation of single-head attention is expressed as: ， in, Indicates Attention head, represent query vector, key vector and value vector respectively, They represent learnable parameters respectively; The outputs of all attention heads are concatenated and the masked multi-head attention output is obtained through linear transformation, which is expressed as: ， in represents a learnable parameter; The output of the masked multi-head attention is added to the respective embedding vectors and normalized. The vector It is expressed as: ， in represents the output of masked multi-head attention, represents the original embedding matrix.

4. A crop seed vitality prediction system based on multi-source information coupling, characterized in that: Used to execute the crop seed vitality prediction method based on multi-source information coupling as described in any one of claims 1 to 3, comprising: a data acquisition unit, a model aggregation unit and an information interaction unit; The data acquisition unit collects multi-source heterogeneous spectral data through hardware equipment, aggregates the multi-source heterogeneous spectral data and parses out seed phenotypic data; The model aggregation unit includes multiple deep learning models, including a SAM large model, a spatiotemporal graph network model, a multi-view diffusion model, a sparse view large-scale reconstruction model, and a seed vitality prediction model constructed by a dual-channel attention aggregation neural network, which extracts the appearance phenotypic data, biochemical parameters, environmental data, and real-time dynamic 3D phenotypic data of seeds based on spectral data and phenotypic data to generate multi-source data of seeds; The information interaction unit includes a background management module and a front-end interaction module. The background management module is responsible for user role information management, log management, and data management; the front-end interaction module is responsible for data visualization display.

Images