Soybean germplasm resource tissue organ characteristic intelligent identification and management method

Abstract

The invention relates to the technical field of computer vision, and discloses a soybean germplasm resource tissue organ feature intelligent identification and management method, which comprises the following steps: building a project management system, creating a soybean germplasm resource project, and inputting project basic information; generating unique identification information, and associating the unique identification information with the corresponding variety basic information; scanning the unique identification information, and collecting image data and text data; preprocessing the image data, making a soybean image data set, and marking the soybean image data set; constructing a soybean tissue organ feature intelligent identification model, and deploying the intelligent identification model to a project management system; performing batch processing by using the model image data to generate an analysis result; and storing the data obtained by intelligent identification to a project management system according to variety resources. According to the method, efficient management and intelligent identification of soybean germplasm resources can be realized, and scientific decision support is provided for soybean breeding and planting.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular to a method for intelligently identifying and managing characteristics of soybean germplasm resources tissues and organs. Background Art

[0002] In modern agricultural production, the management and utilization of soybean germplasm resources are crucial for improving soybean yield and quality. Traditional germplasm management relies primarily on manual record-keeping and simple image acquisition, a method that is both inefficient and error-prone. Furthermore, manual identification of soybean tissue and organ characteristics requires specialized knowledge and experience, making large-scale application difficult. With the advancement of computer vision and machine learning technologies, the application of intelligent recognition systems in agriculture is increasing. However, existing systems primarily focus on pest and disease identification and yield prediction, lacking sufficient support for the identification and management of soybean germplasm tissue and organ characteristics.

[0003] Existing soybean germplasm management systems generally lack the ability to identify and analyze detailed characteristics of soybean tissues and organs. For example, while some systems can record basic soybean information and image data, they are unable to automatically identify and analyze key characteristics such as the number and shape of compound leaflets, flower color, pod hair color, pod curvature, and the number of grains per pod. This results in researchers spending considerable time and effort on manual analysis during germplasm resource evaluation and breeding, severely impacting work efficiency and the accuracy of scientific decision-making.

[0004] Furthermore, existing systems also have shortcomings in data management and analysis. Data storage and management are often fragmented across different platforms and tools, lacking unified management and analysis capabilities. This makes it difficult for researchers to quickly access and analyze large amounts of germplasm data, further limiting the effective utilization of germplasm resources. Therefore, a system is needed that can efficiently identify and manage the tissue and organ characteristics of soybean germplasm resources to improve the efficiency and scientific nature of germplasm resource management.

[0005] Therefore, the present invention proposes a method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics. By building a project management system on a cloud platform, efficient management and intelligent identification of soybean germplasm resources can be achieved, providing scientific decision-making support for soybean breeding and planting. Summary of the Invention

[0006] The purpose of the present invention is to solve the problems of low efficiency in soybean germplasm resource management, inaccurate identification of tissue and organ characteristics, scattered data and lack of unified management in the existing technology, and to propose a method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics, comprising the following steps: Step S1: Building a project management system through the cloud platform, creating a soybean germplasm resource project, and entering basic project information; Step S2: Enter basic information for each soybean variety, generate unique identification information, and associate the unique identification information with the corresponding variety basic information; Step S3, scanning the unique identification information through a mobile device, collecting image data and text data of the corresponding soybean variety, and uploading the collected data to the project management system; Step S4, preprocessing the image data, screening out clear images containing soybean tissue and organ features, creating a soybean image dataset, and annotating the soybean image dataset; Step S5: constructing an intelligent recognition model for soybean tissue and organ characteristics, deploying the intelligent recognition model to a project management system, and adding judgment commands to optimize the output of multi-category detection results; Step S6: Use the trained intelligent recognition model to batch process the collected image data, automatically identify tissue and organ features, and generate analysis results immediately; Step S7: The tissue and organ characteristic data obtained by intelligent identification are saved in the project management system according to the variety resources, and data download and visualization display functions are provided.

[0008] Furthermore, in step S1, the following sub-steps are also included: S1-1, select a cloud service platform, register and log in to a cloud service account. The cloud service platforms include Alibaba Cloud, Tencent Cloud and AWS; S1-2, creating a project management instance on the cloud service platform, installing and deploying a project management system, wherein the project management system supports the creation, management, and data storage functions of soybean germplasm resource projects; S1-3, create a soybean germplasm resource project through the project management system, and enter basic project information, which includes project name, crop name, variety, project leader and list of participants.

[0009] Furthermore, in step S2, the following sub-steps are also included: S2-1, importing basic information of each soybean variety into the project management system using an Excel spreadsheet in a specified format, wherein the basic information of the soybean variety includes variety name, origin, characteristics, cultivation measures, and tissue and organ characteristics; S2-2, automatically identifying the data in the Excel table through the project management system and generating corresponding unique identification information for each variety, wherein the unique identification information includes the variety number, QR code, barcode and RFID tag; S2-3, associating the generated unique identification information with basic information of the soybean variety through the project management system, so that each unique identification information corresponds to one soybean variety; S2-4, the generated unique identification information is printed and made into a ground-insertion plate, which is inserted into the corresponding soybean variety planting plot to complete the allocation of variety information and planting area.

[0010] Furthermore, in step S3, the following sub-steps are also included: S3-1, using a mobile device to scan the unique identification information on the planting sign, and displaying the basic information and historical collection records of the soybean variety corresponding to the unique identification information; S3-2, select the content to be collected. Each content provides two data collection modes: image selection and text entry. The collected content includes compound leaflets, flowers, pods, and other content. The image selection includes images taken by the camera and images imported from the album. The text entry includes entering text information. S3-3, setting an image selection mode, wherein the image selection mode includes shooting with a background and shooting without a background, wherein shooting with a background uses a black light-absorbing cloth as the background; S3-4, setting the criteria for image selection, the criteria for image selection including: When photographing flowers, include at least one complete soybean flower in the frame; When photographing leaflets of compound leaves, the camera frame should include at least one complete leaf, and the leaflets on the leaf should not block each other and remain flat. When photographing pods, at least one pod is facing forward; S3-5, setting the content of text entry, the text entry content includes growing environment, health status, phenological period, yield estimate and other remarks; S3-6, after the collection is completed, click Submit to upload the data to the project management system of the cloud platform, and use cloud storage technology to automatically classify and label the data.

[0011] Furthermore, in step S4, the following sub-steps are also included: S4-1, preprocessing the uploaded image data to select clear images containing soybean tissue and organ features, wherein the preprocessing includes image cropping, normalization, and denoising; S4-2, generating a soybean image dataset based on soybean tissue and organ characteristics, wherein the soybean image dataset includes a dataset of the number of leaflets in a compound leaf, a dataset of the shape of leaflets in a compound leaf, a dataset of flower color, a dataset of the color of pod hairs, a dataset of the degree of pod curvature, and a dataset of the number of grains per pod; S4-3, divide the leaflet number dataset into four categories according to the number of leaflets contained in a single compound leaf: 3 leaves, 4 leaves, 5 leaves, and 6 leaves; S4-4, according to the shape of the leaflets, the compound leaflet shape dataset is divided into six categories: lanceolate, pointed ovate, round ovate, oblong, broad ovate, and round ovate; S4-5, the flower color dataset is divided into two categories: white flowers and purple flowers according to the color of soybean flowers; S4-6, according to the different colors of soybean pod hair, the soybean pod hair color data set is divided into two categories: brown and gray; S4-7, classifying the pod curvature dataset into three categories based on the angles formed by the lines connecting the two vertices of the pod with the center point: no or very weak curvature, medium curvature, and strong curvature. The no or very weak curvature angle is 170°-180°, the medium curvature angle is 130°-169°, and the strong curvature angle is less than 130°. S4-8, according to the number of seeds contained in a single pod, the single pod number dataset is divided into four categories: single pod, two pods, three pods, and four pods; S4-9, Labelimg software is used to label all datasets included in the soybean image dataset according to the division criteria. Each dataset contains images with and without background.

[0012] Furthermore, in step S5, the following sub-steps are also included: S5-1, divide the labeled soybean image dataset into training set, validation set and test set according to the ratio of 70%, 15% and 15% respectively; S5-2, constructing a soybean tissue and organ feature intelligent recognition model, using the YOLOv8 convolutional neural network as the basic architecture of the intelligent recognition model, and training target detection models for the number of soybean leaflets, leaflet shape, flower color, pod hair color, pod curvature, and number of soybean grains per pod; S5-3, configuring the YOLOv8 network and adjusting network parameters to meet the detection requirements of soybean tissue and organ features, wherein the network parameters include input image size, learning rate, batch size, regularization parameter, optimizer, and loss function; S5-4, using the training set to train the YOLOv8 network, and optimizing the network weights through the backpropagation algorithm. The training process includes multiple epochs, and each epoch performs a complete forward and backward propagation on the training set; S5-5, use the validation set to monitor the training process and adjust hyperparameters to improve the performance of the intelligent recognition model and avoid overfitting. The hyperparameters include learning rate, batch size, regularization parameter, dropout rate, learning rate decay strategy, and optimizer type; S5-6, using the test set to evaluate the trained intelligent recognition model, and judging the accuracy and generalization ability of the intelligent recognition model by calculating evaluation indicators, wherein the evaluation indicators include precision, recall rate, and F1 score; S5-7, deploying the intelligent recognition model to the project management system, adding judgment commands based on the intelligent recognition model to optimize the output of multi-category detection results, the judgment commands include: For the classification of leaflet number and leaflet shape, if the object detection model detects multiple categories, the category with the output detection box close to the center of the image pixel is the final result; For flower color and pod fuzz color classification, if the target detection model detects multiple categories, the category with the largest number of outputs is the final result; For the degree of pod curvature and the number of beans in a pod, if the target detection model detects multiple categories, output all categories and the number of beans in each category.

[0013] Furthermore, in step S6, the following sub-steps are also included: S6-1, uploading the collected soybean image data to a project management system, and selecting tissue and organ features through a user interface of the project management system, wherein the tissue and organ features include the number of leaflets of compound leaves, the shape of leaflets of compound leaves, the color of the flower, the color of the pod hairs, the degree of curvature of the pods, and the number of grains per pod; S6-2, calling the deployed soybean tissue and organ feature intelligent recognition model based on the selected tissue and organ features to batch process the uploaded images; S6-3, immediately generating tissue and organ feature analysis results, which include the recognition results of each image, the frequency and distribution of each feature.

[0014] Furthermore, in step S7, the following sub-steps are also included: S7-1, storing the tissue and organ feature data obtained through intelligent identification in a structured manner according to variety resources, and creating a separate database table for each soybean variety. The database table contains all identification results for the variety, and each table has fields including variety number, image number, collection time, and collection location; S7-2, supports data download and visual display through the project management system. The data download includes allowing users to download recognition results and analysis reports. The visual display includes providing a visual interface to display feature distribution maps and statistical charts.

[0015] The beneficial effects brought about by the technical solution provided by the present invention include at least: The present invention is equipped with an intelligent recognition module, which can automatically identify the characteristics of soybean tissue organs through computer vision technology and deep learning algorithms, which not only improves the accuracy and efficiency of recognition, but also reduces the time and cost of manual operation.

[0016] The present invention sets up a data management module, which realizes unified storage and management of data through the cloud platform, facilitates subsequent analysis and retrieval, and enables researchers to quickly obtain and analyze large amounts of germplasm resource data, thereby improving the efficiency and scientificity of data management.

[0017] The present invention sets up a user interface module. Through the user interface of the project management system, researchers can easily select the tissue and organ characteristics of interest, call the intelligent recognition model for batch processing, and instantly generate detailed analysis results. This not only improves the user experience, but also enables non-professional technicians to easily use the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 A flow chart of a method provided by an embodiment of the present invention; Figure 2 Schematic diagram of image shooting with and without background provided by an embodiment of the present invention; Figure 3 A classification diagram of the number of leaflets of compound leaves provided by an embodiment of the present invention; Figure 4 A diagram of the shape classification of compound leaflets provided by an embodiment of the present invention; Figure 5 A color classification diagram provided by an embodiment of the present invention; Figure 6 A classification diagram of bean pod hair colors provided in an embodiment of the present invention; Figure 7 A classification diagram of the degree of curvature of pods provided in an embodiment of the present invention; Figure 8 A classification chart of the number of grains per pod provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail a method for intelligently identifying and managing soybean germplasm tissue and organ characteristics, including its specific implementation, structure, features, and effectiveness. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0021] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0022] The following examples are for illustrative purposes only and are not intended to limit the scope of the present invention.

[0023] The specific scheme of the method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics provided by the present invention is described in detail below with reference to the accompanying drawings.

[0024] See also Figure 1 , which shows a method flow chart of a method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics provided by one embodiment of the present invention, the method comprising the following steps: Step S1: Building a project management system through the cloud platform, creating a soybean germplasm resource project, and entering basic project information; Wherein, step S1 further includes the following sub-steps: S1-1, select a cloud service platform, register and log in to a cloud service account. Cloud service platforms include Alibaba Cloud, Tencent Cloud, and AWS; S1-2, create a project management instance on the cloud service platform, install and deploy the project management system, which supports the creation, management and data storage of soybean germplasm resource projects; S1-3, create a soybean germplasm resource project through the project management system and enter the basic project information, which includes the project name, crop name, variety, project leader and list of participants.

[0025] It should be noted that selecting a cloud service platform: Selecting a reliable cloud service platform is the first step in implementing the present invention. The cloud service platform provides the necessary computing resources, storage resources, and network resources to support the operation of the project management system. Common cloud service platforms include Alibaba Cloud, Tencent Cloud, and AWS. These platforms provide powerful infrastructure and flexible service options to meet the needs of projects of different sizes and needs.

[0026] Register and log in to a cloud service account: Users need to register an account on the selected cloud service platform. The registration process includes providing personal information, setting an account name and password, and verifying the email address or mobile phone number. After registration is completed, users can log in to the cloud service platform and start creating and managing projects.

[0027] Create a project management instance: Create a new project management instance on the cloud service platform. This step involves configuring the required computing resources (CPU, memory), storage resources (hard disk space), and network resources (IP address, bandwidth). Cloud service platforms usually provide a graphical interface or command line tools to facilitate user resource configuration.

[0028] Install and deploy the project management system: The project management system is the core tool for soybean germplasm resource management. The system supports the creation, management and data storage of soybean germplasm resource projects. The installation process includes downloading system software, configuring the database, and setting user permissions. After deployment is complete, the project management system will run on the cloud service platform, and users can access the system through a browser or other client.

[0029] Create a soybean germplasm resource project: Create a new soybean germplasm resource project through the project management system. This step involves filling in the basic information of the project, including project name, project description, and project objectives. The project management system will assign a unique project number to each project to facilitate subsequent management and query.

[0030] Project Name: The name of the project, used to identify and distinguish different projects.

[0031] Crop Name: The name of the crop involved in the soybean germplasm resources to ensure that the project is associated with a specific crop.

[0032] Variety: Information on the soybean varieties involved, including multiple varieties.

[0033] Project Leader: Information about the person responsible for the project, including name and contact information.

[0034] List of participants: List of other personnel involved in the project, including researchers and technicians.

[0035] Step S2: Enter basic information for each soybean variety, generate unique identification information, and associate the unique identification information with the corresponding variety basic information; Wherein, in step S2, the following sub-steps are also included: S2-1, import the basic information of each soybean variety into the project management system using an Excel spreadsheet in a specified format. The basic information of soybean varieties includes variety name, origin, characteristics, cultivation measures, and tissue and organ characteristics; S2-2, automatically identify the data in the Excel table through the project management system and generate corresponding unique identification information for each variety. The unique identification information includes variety number, QR code, barcode and RFID tag; S2-3, associating the generated unique identification information with basic information of the soybean variety through the project management system, so that each unique identification information corresponds to one soybean variety; S2-4, the generated unique identification information is printed and made into a ground-insertion plate, which is inserted into the corresponding soybean variety planting plot to complete the allocation of variety information and planting area.

[0036] It should be noted that an Excel table should be prepared: an Excel table containing basic information of soybean varieties should be prepared. The format of the table should comply with the requirements of the project management system and usually include the following columns: variety name, origin, characteristics, cultivation measures and tissue and organ characteristics.

[0037] Variety names: "Zhonghuang 13", "Heinong 44".

[0038] Origin: The origin or breeding unit of the variety.

[0039] Characteristics: Variety's growth cycle, stress resistance (drought resistance, disease resistance), and yield potential.

[0040] Cultural practices: appropriate planting density, fertilization recommendations, irrigation requirements.

[0041] Tissue and organ characteristics: number of leaflets of compound leaves, shape of leaflets of compound leaves, flower color, color of pod hairs, degree of pod curvature, and number of grains in a single pod.

[0042] Import data: Use the import function of the project management system to import data from Excel spreadsheets into the system in batches. The system will automatically parse the table contents, extract the basic information of each variety, and store it in the database.

[0043] Variety ID: Each soybean variety is assigned a unique numeric ID.

[0044] QR code: A unique QR code is generated for each soybean variety, which contains the variety number and other key information.

[0045] Barcoding: Generate a unique barcode for each soybean variety, which contains the variety number and other key information.

[0046] RFID tag (optional): Generate a unique RFID tag for each soybean variety, which contains the variety number and other key information.

[0047] Association operation: The project management system will associate the generated unique identification information with the basic information of the soybean variety. This step ensures that each unique identification information corresponds to a soybean variety. The system will create an association table to establish a one-to-one correspondence between the unique identification information and the basic information of the variety.

[0048] Print unique identification information: Print the generated unique identification information on the label paper to ensure that the information is clear and readable.

[0049] Make ground-insertion signs: Make the printed label paper into ground-insertion signs. The ground-insertion signs can be made of plastic, wood or other durable materials to ensure long-term use in field environments.

[0050] Insert into the planting plot: Insert the prepared insertion signs into the corresponding soybean variety planting plot. The insertion signs should be placed in a conspicuous position to facilitate researchers and staff to quickly obtain variety information by scanning the unique identification information.

[0051] Complete the allocation: By setting up ground signs, the variety information and planting area are allocated. This step not only facilitates field management, but also provides basic support for subsequent data collection and intelligent identification.

[0052] Step S3, scanning the unique identification information through a mobile device, collecting image data and text data of the corresponding soybean variety, and uploading the collected data to the project management system; Wherein, in step S3, the following sub-steps are also included: S3-1, using a mobile device to scan the unique identification information on the planting sign, and displaying the basic information and historical collection records of the soybean variety corresponding to the unique identification information; S3-2, select the collection content. Each content provides two data collection methods: image selection and text entry. The collection content includes compound leaflets, flowers, pods and other content. Image selection includes camera shots and images imported from the album. Text entry includes entering text information. S3-3, set the image selection method, the image selection method includes shooting with background and shooting without background. Shooting with background uses black light-absorbing cloth as the background; S3-4, set the criteria for image selection, which include: When photographing flowers, include at least one complete soybean flower in the frame; When photographing leaflets of compound leaves, the camera frame should include at least one complete leaf, and the leaflets on the leaf should not block each other and remain flat. When photographing pods, at least one pod is facing forward; S3-5, set the content of text entry, which includes growing environment, health status, phenological period, yield estimate and other notes; S3-6, after the collection is completed, click Submit to upload the data to the project management system of the cloud platform, and use cloud storage technology to automatically classify and label the data.

[0053] Please refer to Figure 2 Schematic diagram of shooting images with and without background provided by an embodiment of the present invention.

[0054] It should be noted that mobile device scanning: use a mobile device (smartphone or tablet) that supports QR code or barcode scanning function, and ensure that the device has installed and configured the dedicated APP or universal scanning tool provided by the project management system.

[0055] Scanning operation: Aim the camera of your mobile device at the unique identification information on the ground insertion sign and scan it. The APP will automatically recognize and parse the identification information.

[0056] Display information: After a successful scan, the app will display the basic information (variety name, origin, characteristics) and historical collection records (previous images and text data) of the soybean variety corresponding to the unique identification information, which provides researchers with instant reference information to ensure the continuity and accuracy of collection work.

[0057] Collection content selection: Researchers select the content to be collected through the APP interface, including leaflets, flowers, pods and other content. Each content provides two data collection methods: image selection and text entry.

[0058] Image selection: Researchers can choose to capture images in real time using the mobile device's camera or import existing images from the mobile device's photo album.

[0059] Text entry: Researchers can enter text information related to the collected content.

[0060] Shooting method selection: Researchers can choose to shoot with or without a background. For shooting with a background, a black light-absorbing cloth is used as the background to reduce the impact of ambient light on image quality and improve image clarity and contrast. For shooting without a background: soybean plants or organs are directly photographed without a background cloth. This method is suitable for rapid acquisition in a natural environment, but subsequent image processing may be required to remove background interference.

[0061] Shooting standards: To ensure the quality and consistency of the collected image data, researchers need to follow the following shooting standards: When photographing flowers: Include at least one complete soybean flower in the frame and ensure the flower structure is clearly visible.

[0062] When photographing the leaflets of compound leaves: the lens frame must include at least one complete compound leaf, and the leaflets on the compound leaf should not block each other and remain flat, so that the number and shape of the leaflets of the compound leaf can be clearly identified.

[0063] When photographing pods: View at least one pod from the front, ensuring the pod's shape and features are clearly visible.

[0064] Growing environment: including soil type, irrigation conditions, and fertilization conditions; Health status: including pest and disease conditions, growth stage; Phenological period: including flowering period and maturity period; Yield estimation: including the number of pods per plant and the number of grains per pod; Other remarks: such as special observation records and abnormal situations; Step S4, preprocessing the image data, screening out clear images containing soybean tissue and organ features, creating a soybean image dataset, and annotating the soybean image dataset; Wherein, in step S4, the following sub-steps are also included: S4-1, preprocessing the uploaded image data to select clear images containing soybean tissue and organ features. The preprocessing includes image cropping, normalization, and denoising; S4-2, based on soybean tissue and organ characteristics, a soybean image dataset is generated. The soybean image dataset includes a dataset of the number of leaflets in a compound leaf, a dataset of the shape of leaflets in a compound leaf, a dataset of flower color, a dataset of the color of pod hairs, a dataset of the degree of pod curvature, and a dataset of the number of grains per pod. S4-3, divide the leaflet number dataset into four categories according to the number of leaflets contained in a single compound leaf: 3 leaves, 4 leaves, 5 leaves, and 6 leaves; Please refer to Figure 3 A classification diagram of the number of leaflets of compound leaves provided by an embodiment of the present invention; S4-4, according to the shape of the leaflets, the compound leaflet shape dataset is divided into six categories: lanceolate, pointed ovate, round ovate, oblong, broad ovate, and round ovate; Please refer to Figure 4 A diagram of the shape classification of compound leaflets provided by an embodiment of the present invention; S4-5, the flower color dataset is divided into two categories: white flowers and purple flowers according to the color of soybean flowers; Please refer to Figure 5 A color classification diagram provided by an embodiment of the present invention; S4-6, according to the different colors of soybean pod hair, the soybean pod hair color data set is divided into two categories: brown and gray; Please refer to Figure 6 A classification diagram of bean pod hair colors provided in an embodiment of the present invention; S4-7, according to the angles formed by the lines connecting the two vertices of the pod with the center point, the pod curvature dataset is divided into three categories: no or very weak curvature, medium curvature, and strong curvature. The angle of no or very weak curvature is 170°-180°, the angle of medium curvature is 130°-169°, and the angle of strong curvature is less than 130°. Please refer to Figure 7 A classification diagram of the degree of curvature of pods provided in an embodiment of the present invention; S4-8, according to the number of seeds contained in a single pod, the single pod number dataset is divided into four categories: single pod, two pods, three pods, and four pods; Please refer to Figure 8 A classification chart of the number of grains per pod provided by an embodiment of the present invention; S4-9, Labelimg software is used to label all datasets included in the soybean image dataset according to the division criteria. Each dataset contains images with and without background.

[0065] It should be noted that image cropping: in order to remove irrelevant parts in the image and improve the efficiency and accuracy of subsequent processing, the uploaded image is cropped. The cropping range can be adjusted according to the position of the soybean plant or organ to ensure that the cropped image only contains the part of interest.

[0066] Normalization: Normalizes the pixel values of an image to a fixed range (0 to 1) to eliminate differences in pixel values between different images due to lighting conditions and shooting equipment. Normalization helps improve the generalization ability of intelligent recognition models for images.

[0067] Denoising: Use image denoising algorithms (median filtering, Gaussian filtering) to remove noise from the image and improve image clarity and quality. Denoising can reduce the impact of noise on subsequent feature extraction and recognition.

[0068] Compound leaflet number dataset: contains images of compound leaves with different numbers of leaflets.

[0069] Compound Leaflet Shape Dataset: Contains images of different compound leaflet shapes.

[0070] Flower Color Dataset: Contains images of different flower colors.

[0071] Bean Pod Hair Color Dataset: Contains images of different bean pod hair colors.

[0072] Peapod curvature dataset: Contains images of peapods with different curvature degrees.

[0073] Single pod grain number dataset: contains images with different single pod grain numbers.

[0074] Classification criteria for the number of leaflets in compound leaves: The leaflet number dataset is divided into four categories based on the number of leaflets contained in a single compound leaf: 3-leaf: Compound leaves contain 3 leaflets.

[0075] 4-leaf: Compound leaves contain 4 leaflets.

[0076] 5-leaf: Compound leaves contain 5 leaflets.

[0077] 6-leaf: Compound leaves contain 6 leaflets.

[0078] Compound leaflet number classification method: Classify the number of compound leaflets in each image through manual annotation or automatic recognition algorithm, and assign the image to the corresponding category.

[0079] Compound leaflet shape classification criteria: The compound leaflet shape dataset is divided into six categories based on the shape of the leaflets: Lanceolate: The leaf tip is sharp, the leaf base is wide, and the length-to-width ratio of the circumscribed rectangle is r≥2.5; Pointed ovate: The leaf tip is sharp, the leaf base is wide, and the length-to-width ratio of the circumscribed rectangle is 1.5≤r<2.5; Oval: The leaf tip is sharp, the leaf base is wide and rounded, and the aspect ratio of the circumscribed rectangle is r<1.5; Oblong: The leaf tip is sharp, the leaf base is narrow, and the length-to-width ratio of the circumscribed rectangle is r≥2.5; Broad oval: The leaf tip is sharp, the leaf base is narrow, and the length-to-width ratio of the circumscribed rectangle is 1.5≤r<2.5; Round-oval: The leaf tip is shield-shaped, the leaf base is wide but not rounded, and the aspect ratio of the circumscribed rectangle is r<1.5.

[0080] Compound leaflet shape classification method: The compound leaflet shapes in each image are classified through manual annotation or automatic recognition algorithm, and the images are assigned to the corresponding categories.

[0081] Flower color classification standard: According to the color of soybean flowers, the flower color dataset is divided into two categories: White flowers: The flower color is white.

[0082] Purple flower: The flower color is purple.

[0083] Flower color classification method: Classify the flower colors in each image through manual labeling or automatic recognition algorithm, and assign the image to the corresponding category.

[0084] Bean pod fuzz color classification standard: The pod fuzz color dataset is divided into two categories based on the color of the pod fuzz: Brown: The undercoat is brown in color.

[0085] Gray: The undercoat is gray.

[0086] Bean pod fuzz color classification method: Classify the pod fuzz color in each image through manual annotation or automatic recognition algorithm, and assign the image to the corresponding category.

[0087] Pod curvature classification criteria: The pod curvature dataset is divided into three categories based on the angle formed by the line connecting the two vertices of the pod and the center point: No or very weak bend: angle is 170°-180°.

[0088] Medium bend: Angle is 130°-169°.

[0089] Strong bending degree: angle less than 130°.

[0090] Pod curvature classification method: The degree of pod curvature in each image is classified through manual annotation or automatic recognition algorithm, and the image is assigned to the corresponding category.

[0091] Classification criteria for single pod number: The single pod number dataset is divided into four categories based on the number of seeds contained in a single pod: Single pod: A pod containing one seed.

[0092] Two-pod: A pod containing 2 seeds.

[0093] Triple pod: A pod containing 3 seeds.

[0094] Quad Pods: Pods contain 4 seeds.

[0095] Single pod grain number classification method: Classify the single pod grain number in each image through manual annotation or automatic recognition algorithm, and assign the image to the corresponding category.

[0096] Labeling tools: Use Labelimg software to label the images in each dataset. Labelimg is an open source image annotation tool that supports multiple annotation formats, including PASCAL VOC and YOLO.

[0097] Annotation content: According to the classification criteria, the soybean tissue and organ features in the image are annotated. The annotation content includes the feature category and location (bounding box coordinates) information.

[0098] Annotation type: Each dataset contains images with and without background. Images with background are used to improve the intelligent recognition model's ability to recognize features, while images without background are used to test the robustness of the intelligent recognition model in complex environments.

[0099] Step S5: constructing an intelligent recognition model for soybean tissue and organ characteristics, deploying the intelligent recognition model to a project management system, and adding judgment commands to optimize the output of multi-category detection results; Wherein, in step S5, the following sub-steps are also included: S5-1, divide the labeled soybean image dataset into training set, validation set and test set according to the ratio of 70%, 15% and 15% respectively; S5-2: Build an intelligent recognition model for soybean tissue and organ features. The intelligent recognition model uses the YOLOv8 convolutional neural network as its infrastructure. It trains target detection models for the number of leaflets, leaflet shape, flower color, pod hair color, pod curvature, and number of grains per pod. S5-3, configure the YOLOv8 network and adjust the network parameters to meet the detection requirements of soybean tissue and organ features. The network parameters include input image size, learning rate, batch size, regularization parameter, optimizer, and loss function; S5-4, use the training set to train the YOLOv8 network, and optimize the network weights through the backpropagation algorithm. The training process includes multiple epochs, and each epoch performs a complete forward and backward propagation on the training set; S5-5, use the validation set to monitor the training process and adjust hyperparameters to improve the performance of the intelligent recognition model and avoid overfitting. Hyperparameters include learning rate, batch size, regularization parameter, dropout rate, learning rate decay strategy, and optimizer type; S5-6, use the test set to evaluate the trained intelligent recognition model, and judge the accuracy and generalization ability of the intelligent recognition model by calculating the evaluation indicators, including precision, recall rate and F1 score; S5-7, deploy the intelligent recognition model to the project management system. Add judgment commands based on the intelligent recognition model to optimize the output of multi-category detection results. The judgment commands include: For the classification of leaflet number and leaflet shape, if the object detection model detects multiple categories, the category with the output detection box close to the center of the image pixel is the final result; For flower color and pod fuzz color classification, if the target detection model detects multiple categories, the category with the largest number of outputs is the final result; For the degree of pod curvature and the number of beans in a pod, if the target detection model detects multiple categories, output all categories and the number of beans in each category.

[0100] It should be noted that the dataset is divided into three parts: training set, validation set and test set, with a ratio of 70%, 15% and 15% respectively, in order to ensure the effectiveness and reliability of intelligent recognition model training.

[0101] Training set: used to train the intelligent recognition model, accounting for 70% of the total data set.

[0102] Validation set: used to monitor the training process, adjust hyperparameters, and avoid overfitting, accounting for 15% of the total dataset.

[0103] Test set: used to evaluate the final performance of the intelligent recognition model, accounting for 15% of the total dataset.

[0104] Constructing an intelligent recognition model for soybean tissue and organ characteristics: The intelligent recognition model uses the YOLOv8 convolutional neural network as its basic architecture, and trains target detection models for soybean tissue and organ characteristics (the number of leaflets in soybean compound leaves, the shape of leaflets in compound leaves, the color of flower pod hairs, the degree of pod curvature, and the number of grains in a single pod). The intelligent recognition model is Choose YOLOv8: YOLOv8 is an advanced convolutional neural network architecture suitable for real-time target detection tasks. It strikes a good balance between speed and accuracy and is suitable for detecting soybean tissue and organ features.

[0105] Intelligent recognition model training: Six target detection models are trained for six soybean tissue and organ characteristics (number of leaflets, shape of leaflets, flower color, pod hair color, pod curvature, and number of grains per pod). Each target detection model focuses on identifying a specific feature. The intelligent recognition model is the sum of the six target detection models, using the YOLOv8 convolutional neural network as its infrastructure. This improves the accuracy and robustness of the intelligent recognition model.

[0106] Input Image Size: Adjust the resolution of the input image to accommodate image data of different sizes.

[0107] Learning rate: Set the initial learning rate and adjust the learning rate according to the training progress to optimize the training process.

[0108] Batch size: Set the number of images per training batch to balance training speed and memory usage.

[0109] Regularization parameter: Set the regularization parameter of weight decay to prevent overfitting.

[0110] Optimizer: Select an appropriate optimizer (Adam, SGD) and adjust its parameters.

[0111] Loss function: Select an appropriate loss function (cross entropy loss, IoU loss) and adjust its weight.

[0112] Training process: The YOLOv8 network is trained using the training set. The training process includes multiple epochs, and each epoch performs a complete forward and backward propagation on the training set.

[0113] Forward propagation: Input the images in the training set into the network and calculate the output of the network.

[0114] Loss calculation: Calculate the loss value between the network output and the true label to evaluate the prediction results of the intelligent recognition model.

[0115] Back propagation: The gradient of the loss function with respect to the network parameters is calculated through the back propagation algorithm.

[0116] Weight update: Update the network weights according to the calculated gradients and optimize the network parameters.

[0117] Monitor the training process: Use the validation set to monitor the training process and evaluate the performance of the intelligent recognition model on unseen data.

[0118] Hyperparameter Tuning: Based on the performance of the validation set, adjust the hyperparameters to improve the performance of the intelligent recognition model. Key hyperparameters include: Learning rate: Adjust the learning rate based on the performance of the validation set, usually using a learning rate decay strategy.

[0119] Batch size: Adjust the batch size to balance training speed and smart recognition model performance.

[0120] Regularization parameter: Adjust the regularization parameter to prevent overfitting.

[0121] Dropout rate: Adjust the Dropout rate to reduce co-adaptation between neurons and improve the generalization ability of the intelligent recognition model.

[0122] Learning rate decay strategy: Adopt an appropriate learning rate decay strategy (decay by 0.1 times every 10 epochs) to optimize the training process.

[0123] Optimizer Type: Select an appropriate optimizer (Adam or SGD) to speed up the training process.

[0124] Intelligent recognition model evaluation: Use the test set to evaluate the trained intelligent recognition model to ensure the performance of the intelligent recognition model on unseen data.

[0125] Evaluation metrics: Calculate evaluation metrics to determine the accuracy and generalization ability of the intelligent recognition model. The evaluation metrics include: Precision: The proportion of samples predicted as positive by the intelligent recognition model that are actually positive.

[0126] Recall: The ratio of samples that are actually positive that are predicted as positive by the intelligent recognition model.

[0127] F1 score: The harmonic mean of precision and recall, which comprehensively evaluates the performance of the intelligent recognition model.

[0128] Intelligent recognition model deployment: Deploy the six trained object detection models to the project management system to ensure that the intelligent recognition model can run efficiently in the system.

[0129] Judgment command: Add judgment commands based on the intelligent recognition model to optimize the output of multi-category detection results. Specific judgment commands include: Classification of the number and shape of compound leaves: If the object detection model detects multiple categories, the category with the output detection box close to the center of the image pixel is the final result.

[0130] Flower color and pod fuzz color classification: If the target detection model detects multiple categories, the category with the largest number of outputs is the final result.

[0131] Pod curvature and number of beans per pod: If the target detection model detects multiple categories, output all categories and the number of beans in each category.

[0132] Step S6: Use the trained intelligent recognition model to batch process the collected image data, automatically identify tissue and organ features, and generate analysis results immediately; Wherein, in step S6, the following sub-steps are also included: S6-1, uploading the collected soybean image data to the project management system, and selecting tissue and organ features through the user interface of the project management system, the tissue and organ features including the number of leaflets of compound leaves, the shape of leaflets of compound leaves, the color of the flower, the color of the pod hairs, the degree of pod curvature, and the number of grains per pod; S6-2, calling the deployed soybean tissue and organ feature intelligent recognition model based on the selected tissue and organ features to batch process the uploaded images; S6-3, instantly generate tissue and organ feature analysis results, including the recognition results of each image, the frequency and distribution of each feature.

[0133] It should be noted that image data uploading: researchers upload the collected soybean image data to the project management system through mobile devices or computers. The uploading process can be completed through network transmission to ensure the security and integrity of the data.

[0134] User interface selection: The project management system provides an intuitive user interface through which researchers can select the tissue and organ characteristics that need to be identified. The optional tissue and organ characteristics include the number of leaflets in compound leaves, the shape of leaflets in compound leaves, the color of flower bud hairs, the degree of pod curvature, and the number of grains per pod.

[0135] Feature selection: Researchers select one or more tissue and organ features based on research needs. The selected features will determine the type of intelligent recognition model that will be called subsequently.

[0136] Intelligent recognition model call: The project management system automatically calls the corresponding intelligent recognition model based on the tissue and organ features selected by the user. Each feature has a specially trained target detection model to ensure the accuracy and efficiency of recognition.

[0137] Batch processing: The system processes the uploaded image data in batches. Each image is sequentially identified by the selected intelligent recognition model. Batch processing can significantly improve processing efficiency and is suitable for rapid analysis of large amounts of image data.

[0138] Processing flow: 1. Image preprocessing: Perform necessary preprocessing on the uploaded images, including adjusting the image size and normalization, to meet the input requirements of the intelligent recognition model.

[0139] 2. Feature extraction: The intelligent recognition model extracts key features from the preprocessed image.

[0140] 3. Feature recognition: The intelligent recognition model performs recognition based on the extracted features and outputs the recognition results.

[0141] Generate results instantly: After completing batch processing, the intelligent recognition model generates analysis results instantly. The result generation process is usually completed within a few seconds to a few minutes, depending on the number of images and model complexity.

[0142] Recognition results for each image: The system generates a detailed recognition report for each image, including the recognized tissue and organ features and their location information.

[0143] Feature occurrence frequency: The system counts the number of times each feature appears in all images and calculates the frequency of occurrence.

[0144] Distribution: The system analyzes the distribution of each feature in different images and generates a distribution graph or statistical table.

[0145] Step S7: The tissue and organ characteristic data obtained by intelligent identification are saved in the project management system according to the variety resources, and data download and visualization display functions are provided; Wherein, in step S7, the following sub-steps are also included: S7-1: The tissue and organ feature data obtained by intelligent identification are stored in a structured manner according to variety resources. A separate database table is created for each soybean variety. The database table contains all the identification results of the variety. The fields of each table include variety number, image number, collection time and collection location; S7-2 supports data download and visualization through the project management system. Data download includes allowing users to download recognition results and analysis reports. Visualization includes providing a visualization interface to display feature distribution graphs and statistical charts.

[0146] It should be noted that structured storage: In order to facilitate management and query, the tissue and organ characteristic data obtained by intelligent identification needs to be stored in a structured manner. Structured storage refers to storing data in a database with a clear structure and format to facilitate subsequent query, analysis and processing.

[0147] Database table design: Create a separate database table for each soybean variety. The fields of each table are designed as follows: Variety number: A number that uniquely identifies each soybean variety.

[0148] Image number: A number that uniquely identifies each image.

[0149] Acquisition time: The specific timestamp of image acquisition.

[0150] Collection location: The specific location where the image was collected.

[0151] Identification results: The tissue and organ feature results output by the intelligent recognition model include the number of leaflets, the shape of leaflets, flower color, pod hair color, pod curvature, and the number of grains per pod.

[0152] Data download function: The project management system provides a data download function, allowing users to download identification results and analysis reports. The download content includes: Recognition results: tissue and organ feature recognition results for each image.

[0153] Analysis report: A report containing statistical information about the frequency and distribution of features.

[0154] Visual display function: The project management system provides an intuitive visual interface to display feature distribution graphs and statistical charts. The visual interface includes: Feature distribution graph: Displays the distribution of each feature in different images in the form of a graph, including a bar graph and a line graph.

[0155] Statistical charts: Display statistical information of features, including frequency, mean, and standard deviation.

[0156] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics, characterized in that: The method includes: Step S1: Building a project management system through the cloud platform, creating a soybean germplasm resource project, and entering basic project information; Step S2: Enter basic information for each soybean variety, generate unique identification information, and associate the unique identification information with the corresponding variety basic information; Step S3, scanning the unique identification information through a mobile device, collecting image data and text data of the corresponding soybean variety, and uploading the collected data to the project management system; Step S4, preprocessing the image data, screening out clear images containing soybean tissue and organ features, creating a soybean image dataset, and annotating the soybean image dataset; Step S5: constructing an intelligent recognition model for soybean tissue and organ characteristics, deploying the intelligent recognition model to a project management system, and adding judgment commands to optimize the output of multi-category detection results; Step S6: Use the trained intelligent recognition model to batch process the collected image data, automatically identify tissue and organ features, and generate analysis results immediately; Step S7: The tissue and organ characteristic data obtained by intelligent identification are saved in the project management system according to the variety resources, and data download and visualization display functions are provided.

2. The method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics according to claim 1, characterized in that: Wherein, step S1 further includes the following sub-steps: S1-1, select a cloud service platform, register and log in to a cloud service account. The cloud service platforms include Alibaba Cloud, Tencent Cloud and AWS; S1-2, creating a project management instance on the cloud service platform, installing and deploying a project management system, wherein the project management system supports the creation, management, and data storage functions of soybean germplasm resource projects; S1-3, create a soybean germplasm resource project through the project management system, and enter basic project information, which includes project name, crop name, variety, project leader and list of participants.

3. The method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics according to claim 1, characterized in that: Wherein, in step S2, the following sub-steps are also included: S2-1, importing basic information of each soybean variety into the project management system using an Excel spreadsheet in a specified format, wherein the basic information of the soybean variety includes variety name, origin, characteristics, cultivation measures, and tissue and organ characteristics; S2-2, automatically identifying the data in the Excel table through the project management system and generating corresponding unique identification information for each variety, wherein the unique identification information includes the variety number, QR code, barcode and RFID tag; S2-3, associating the generated unique identification information with basic information of the soybean variety through the project management system, so that each unique identification information corresponds to one soybean variety; S2-4, the generated unique identification information is printed and made into a ground-insertion plate, which is inserted into the corresponding soybean variety planting plot to complete the allocation of variety information and planting area.

4. The method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics according to claim 1, characterized in that: Wherein, in step S3, the following sub-steps are also included: S3-1, using a mobile device to scan the unique identification information on the planting sign, and displaying the basic information and historical collection records of the soybean variety corresponding to the unique identification information; S3-2, select the content to be collected. Each content provides two data collection modes: image selection and text entry. The collected content includes compound leaflets, flowers, pods, and other content. The image selection includes images taken by the camera and images imported from the album. The text entry includes entering text information. S3-3, setting an image selection mode, wherein the image selection mode includes shooting with a background and shooting without a background, wherein shooting with a background uses a black light-absorbing cloth as the background; S3-4, setting the criteria for image selection, the criteria for image selection including: When photographing flowers, include at least one complete soybean flower in the frame; When photographing leaflets of compound leaves, the camera frame should include at least one complete leaf, and the leaflets on the leaf should not block each other and remain flat. When photographing pods, at least one pod is facing forward; S3-5, setting the content of text entry, the text entry content includes growing environment, health status, phenological period, yield estimate and other remarks; S3-6, after the collection is completed, click Submit to upload the data to the project management system of the cloud platform, and use cloud storage technology to automatically classify and label the data.

5. The method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics according to claim 1, characterized in that: Wherein, in step S4, the following sub-steps are also included: S4-1, preprocessing the uploaded image data to select clear images containing soybean tissue and organ features, wherein the preprocessing includes image cropping, normalization, and denoising; S4-2, generating a soybean image dataset based on soybean tissue and organ characteristics, wherein the soybean image dataset includes a dataset of the number of leaflets in a compound leaf, a dataset of the shape of leaflets in a compound leaf, a dataset of flower color, a dataset of the color of pod hairs, a dataset of the degree of pod curvature, and a dataset of the number of grains per pod; S4-3, divide the leaflet number dataset into four categories according to the number of leaflets contained in a single compound leaf: 3 leaves, 4 leaves, 5 leaves, and 6 leaves; S4-4, according to the shape of the leaflets, the compound leaflet shape dataset is divided into six categories: lanceolate, pointed ovate, round ovate, oblong, broad ovate, and round ovate; S4-5, the flower color dataset is divided into two categories: white flowers and purple flowers according to the color of soybean flowers; S4-6, according to the different colors of soybean pod hair, the soybean pod hair color data set is divided into two categories: brown and gray; S4-7, classifying the pod curvature dataset into three categories based on the angles formed by the lines connecting the two vertices of the pod with the center point: no or very weak curvature, medium curvature, and strong curvature. The no or very weak curvature angle is 170°-180°, the medium curvature angle is 130°-169°, and the strong curvature angle is less than 130°. S4-8, according to the number of seeds contained in a single pod, the single pod number dataset is divided into four categories: single pod, two pods, three pods, and four pods; S4-9, Labelimg software is used to label all datasets included in the soybean image dataset according to the division criteria. Each dataset contains images with and without background.

6. The method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics according to claim 1, characterized in that: Wherein, in step S5, the following sub-steps are also included: S5-1, divide the labeled soybean image dataset into training set, validation set and test set according to the ratio of 70%, 15% and 15% respectively; S5-2, constructing a soybean tissue and organ feature intelligent recognition model, using the YOLOv8 convolutional neural network as the basic architecture of the intelligent recognition model, and training target detection models for the number of soybean leaflets, leaflet shape, flower color, pod hair color, pod curvature, and number of soybean grains per pod; S5-3, configuring the YOLOv8 network and adjusting network parameters to meet the detection requirements of soybean tissue and organ features, wherein the network parameters include input image size, learning rate, batch size, regularization parameter, optimizer, and loss function; S5-4, using the training set to train the YOLOv8 network, and optimizing the network weights through the backpropagation algorithm. The training process includes multiple epochs, and each epoch performs a complete forward and backward propagation on the training set; S5-5, use the validation set to monitor the training process and adjust hyperparameters to improve the performance of the intelligent recognition model and avoid overfitting. The hyperparameters include learning rate, batch size, regularization parameter, dropout rate, learning rate decay strategy, and optimizer type; S5-6, using the test set to evaluate the trained intelligent recognition model, and judging the accuracy and generalization ability of the intelligent recognition model by calculating evaluation indicators, wherein the evaluation indicators include precision, recall rate, and F1 score; S5-7, deploying the intelligent recognition model to the project management system, adding judgment commands based on the intelligent recognition model to optimize the output of multi-category detection results, the judgment commands include: For the classification of leaflet number and leaflet shape, if the object detection model detects multiple categories, the category with the output detection box close to the center of the image pixel is the final result; For flower color and pod fuzz color classification, if the target detection model detects multiple categories, the category with the largest number of outputs is the final result; For the degree of pod curvature and the number of beans in a pod, if the target detection model detects multiple categories, output all categories and the number of beans in each category.

7. The method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics according to claim 1, characterized in that: Wherein, in step S6, the following sub-steps are also included: S6-1, uploading the collected soybean image data to a project management system, and selecting tissue and organ features through a user interface of the project management system, wherein the tissue and organ features include the number of leaflets of compound leaves, the shape of leaflets of compound leaves, the color of the flower, the color of the pod hairs, the degree of curvature of the pods, and the number of grains per pod; S6-2, calling the deployed soybean tissue and organ feature intelligent recognition model based on the selected tissue and organ features to batch process the uploaded images; S6-3, immediately generating tissue and organ feature analysis results, which include the recognition results of each image, the frequency and distribution of each feature.

8. The method for intelligent identification and management of soybean germplasm resource tissue and organ characteristics according to claim 1, characterized in that: Wherein, in step S7, the following sub-steps are also included: S7-1, storing the tissue and organ feature data obtained through intelligent identification in a structured manner according to variety resources, and creating a separate database table for each soybean variety. The database table contains all identification results for the variety, and each table has fields including variety number, image number, collection time, and collection location; S7-2, supports data download and visual display through the project management system. The data download includes allowing users to download recognition results and analysis reports. The visual display includes providing a visual interface to display feature distribution maps and statistical charts.

Images