Paddy rice panicle grain phenotype extraction method, system, medium and equipment
By segmenting and identifying rice ear grain images using YOLOV8 and YOLOX models, the problem of low phenotype acquisition efficiency in the prior art is solved, and efficient and accurate phenotype acquisition and analysis are achieved.
Patent Information
- Application Number
- CN202411800459.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-09
- Publication Date
- 2025-04-01
AI Technical Summary
In the prior art, the phenotype acquisition efficiency of rice rice ears is low, and it is difficult to achieve high-throughput phenotype characteristics acquisition.
Multiple instance segmentation models were used to segment the rice ear images, and the YOLOV8 model with the highest segmentation accuracy was selected as the rice ear image segmentation model, and the classified rice ear images were characterized by convolutional neural network, and the YOLOX network with the highest recognition accuracy was selected as the ear grain recognition model to obtain the rice ear grain-related phenotype.
It realizes efficient acquisition of rice ear phenotypes, improves the efficiency of phenotype acquisition, and can more accurately identify and analyze the status and characteristics of rice ears.
Smart Images
Figure CN119741687A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of agronomy, and in particular to a method, system, medium and equipment for extracting rice spikelet phenotypes. Background Art
[0002] Rice is a species of herbaceous rice, and is the most important and oldest species in the genus Oryza as a grain. It is also called Asian cultivated rice, or simply rice. The flowers of rice are very small. When blooming, the main flower branches will be arched, and small flowers will bloom 30 to 50 cm below the branches. Most of them are self-pollinated and produce seeds, called rice ears. The number of rice grains per ear is a key trait that affects grain cultivation, management and subsequent yields.
[0003] In the rice breeding process, a large number of relevant phenotypic traits are needed as the basis for gene selection, which are directly related to yield and quality traits. Usually, the phenotypic characteristics of rice ears are understood by directly observing the morphological characteristics of rice ears, or by using measuring tools to measure the specific size and weight of rice ears. However, the above methods for determining the phenotypic characteristics of rice ears cannot be performed in a high-throughput manner, which leads to low efficiency in obtaining rice ear phenotypes. Summary of the invention
[0004] The present invention provides a method for extracting rice panicle and grain phenotypes, which is used to solve the above-mentioned problem existing in the prior art, that is, how to improve the efficiency of obtaining rice panicle phenotypes in the prior art. The present invention provides a method for extracting rice panicle and grain phenotypes, which comprises:
[0005] Acquire the rice ear image of the area to be detected;
[0006] Multiple instance segmentation models are used to segment the rice ear images. The YOLOv8 model with the highest segmentation accuracy is selected as the rice ear image segmentation model. The rice ear images are input into the rice ear image segmentation model for segmentation. The segmented rice ear images are classified and labeled according to half-grain, full-grain and full-grain, and the classified labeling data is determined.
[0007] According to the classified labeled data, multiple convolutional neural networks are used to identify the features of the classified rice ear images. Based on the comparative analysis of the recognition results of each target detection network, the YOLOX network with the highest recognition accuracy is used as the ear-grain recognition model.
[0008] The segmented rice ear image is input into the ear-grain recognition model to determine the state of the rice ear and grain, and the rice ear and grain in the classified and labeled rice ear image are recognized to obtain the recognition result, and the rice ear and grain related phenotype is obtained according to the recognition result.
[0009] Optionally, the segmented rice ear images are classified and labeled according to half-grain, full-grain and full-grain, specifically including:
[0010] The labeling tool Labelme was used to classify and label the segmented rice ear images into half-grain, full-grain and full-grain empty grains.
[0011] Optionally, after determining the state of the rice ears and grains, the rice seed setting rate is obtained using the following formula:
[0012]
[0013] Among them, RSSR a is the rice fruit setting rate, NF is the number of whole grains per ear, NE is the number of whole and empty grains per ear, NH is the number of half grains per ear, and PH is the proportion of whole grains per ear.
[0014] Optionally, the rice panicle-grain related phenotypes specifically include:
[0015] Color characteristics, texture characteristics, rice panicle aspect ratio and morphological characteristics.
[0016] Optionally, the parameters of the YOLOV8 model are: Epoch is set to 300, and batch size is set to 4.
[0017] Also includes:
[0018] Before using multiple convolutional neural networks to perform feature recognition on the classified rice ear images, data enhancement processing is performed on the classified and labeled data.
[0019] The present invention provides a rice spike and grain phenotype extraction system, comprising:
[0020] An acquisition module is used to acquire the rice ear image of the area to be detected;
[0021] A labeling module is used to segment the rice ear image using multiple instance segmentation models, select the YOLOv8 model with the highest segmentation accuracy as the rice ear image segmentation model, input the rice ear image into the rice ear image segmentation model for segmentation, classify and label the segmented rice ear image according to half-grain, full-grain and full-grain, and determine the classification labeling data;
[0022] The recognition module is used to perform feature recognition on the classified rice ear images using multiple convolutional neural networks based on the classified labeled data, and to compare and analyze the recognition results of each target detection network, and to use the YOLOX network with the highest recognition accuracy as the ear-grain recognition model;
[0023] The determination module is used to input the segmented rice ear image into the ear-grain recognition model, determine the state of the rice ear and grain, identify the rice ear and grain in the classified and labeled rice ear image, obtain the recognition result, and obtain the rice ear and grain related phenotype according to the recognition result.
[0024] The present invention provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the rice spikelet phenotype extraction method is implemented.
[0025] The present invention provides a computer device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the rice spikelet phenotype extraction method when executing the program.
[0026] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention provides a method for extracting rice ear grain phenotypes, which segments rice ear images through multiple instance segmentation models, and selects the YOLOV8 network with the highest segmentation accuracy as the rice ear image segmentation model, which can better adapt to rice ears of different scales, thereby realizing accurate segmentation of rice ears; by comparing and selecting the YOLOX network with the highest recognition accuracy as the ear-grain recognition model, the segmented rice ear image is input into the ear-grain recognition model to obtain rice ear related phenotypes, thereby effectively improving the efficiency of obtaining rice ear phenotypes. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.
[0028] Figure 1 A flow chart of a rice spike phenotype extraction method provided in an embodiment of the present invention;
[0029] Figure 2 A schematic diagram of obtaining rice fruit setting rate using the target detection model YOLOX provided in an embodiment of the present invention;
[0030] Figure 3 A network structure diagram of the instance segmentation model YOLOV8 provided in an embodiment of the present invention;
[0031] Figure 4 A network structure diagram of the target detection model YOLOX provided in an embodiment of the present invention;
[0032] Figure 5 The running time of obtaining the fruit setting rate of each rice plant by using the target detection model YOLOX provided in the embodiment of the present invention;
[0033] Figure 6 This is a functional framework diagram of a rice spike and grain phenotype extraction system provided by an embodiment of the present invention;
[0034] Figure 7 This is a diagram of a user registration interface of a rice spike-grain phenotype extraction system provided by an embodiment of the present invention;
[0035] Figure 8 This is a user login interface diagram of a rice spike-grain phenotype extraction system provided by an embodiment of the present invention;
[0036] Fig. 9 This is an interface diagram of a rice panicle phenotype analysis module of a rice panicle-grain phenotype extraction system provided by an embodiment of the present invention;
[0037] Fig.10 This is a single analysis selection interface diagram of a rice spike and grain phenotype extraction system provided by an embodiment of the present invention;
[0038] Fig.11 This is a diagram of an operation interface for uploading a picture of a rice spikelet phenotype extraction system provided by an embodiment of the present invention;
[0039] Fig.12 It is a schematic diagram showing a single phenotypic result of the front end of the rice spike and grain phenotypic extraction system provided by an embodiment of the present invention;
[0040] Fig.13 It is a schematic diagram showing a single phenotypic result of the back end of the rice spike and grain phenotype extraction system provided by an embodiment of the present invention;
[0041] Fig.14 It is a single analysis data export diagram of the rice spike and grain phenotype extraction system provided by an embodiment of the present invention;
[0042] Fig.15 It is a plurality of analysis data export diagrams of the rice spike and grain phenotype extraction system provided by an embodiment of the present invention;
[0043] Fig.16 It is a schematic diagram of a single data visualization result of the rice spike and grain phenotype extraction system provided by an embodiment of the present invention;
[0044] Fig.17 It is a schematic diagram of multiple data visualization results of the rice panicle and grain phenotype extraction system provided in an embodiment of the present invention.
[0045] Fig.18 A schematic diagram of a computer device for a rice panicle and grain phenotype extraction method provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0046] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0047] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present invention will be described below in conjunction with the accompanying drawings.
[0048] like Figure 1 and Figure 2 As shown, this embodiment shows a method for extracting rice spikelet phenotypes, comprising:
[0049] S1: Obtain the rice ear image in the area to be detected.
[0050] S2: Use multiple instance segmentation models to segment the rice ear image, select the YOLOv8 model with the highest segmentation accuracy as the rice ear image segmentation model, input the rice ear image into the rice ear image segmentation model for segmentation, classify and label the segmented rice ear image according to half-grain, full-grain and full-grain, and determine the classified labeling data.
[0051] The evaluation indicators of the YOLOV8 model are compared with those of the Mask R-CNN, YOLACT, YOLOV5 and YOLOV7 models, as shown in Table 1.
[0052] Table 1 Evaluation indicators of different instance segmentation models
[0053]
[0054] As shown in Table 1, the YOLOV8 model can better adapt to targets of different scales and achieve more accurate instance segmentation. Therefore, the YOLOV8 model is selected for the segmentation of rice spike images. Figure 3 shown.
[0055] For example, the parameters of the YOLOV8 model can be set to 300 Epochs and 4 batch sizes, and the model accuracy reaches 94.30%.
[0056] For example, when segmenting a rice ear image, it was found through repeated experiments that segmenting a single rice ear image into 4×6=24 blocks has the best recognition effect. Therefore, the rice ear image can be input into a rice ear image segmentation model and divided into 24 blocks before the image is classified and labeled.
[0057] S3: Based on the classified labeled data, multiple convolutional neural networks are used to perform feature recognition on the classified rice ear images. Based on the comparative analysis of the recognition results of each target detection network, the YOLOX network with the highest recognition accuracy is used as the ear-grain recognition model.
[0058] Exemplarily, multiple convolutional neural networks include SD, FastR-CNN, YOLOX, YOLOV7, and YOLOV8. Multiple convolutional neural networks are used to extract and train the classified rice ear images to obtain the recognition effect diagrams and basic networks of each model, as shown in Table 2.
[0059] Table 2 Evaluation indicators of different target detection models
[0060]
[0061] As shown in Table 2, the YOLOX model can achieve more accurate instance segmentation. Therefore, the YOLOX convolutional neural network with the highest recognition accuracy is used to identify the ear grains in the rice ear image. Figure 4 As shown, the rice ear images are continuously extracted, recognized and trained through the YOLOX model, the training set map is 98.10%, and the ear-grain recognition model is obtained.
[0062] For example, after determining the state of the rice ears and grains, the rice grain setting rate can be obtained by using the following formula:
[0063]
[0064] Among them, RSSR a is the rice fruit setting rate, NF is the number of whole grains per ear, NE is the number of whole and empty grains per ear, NH is the number of half grains per ear, and PH is the proportion of whole grains per ear.
[0065] like Figure 5 As shown, the average running time for inference of the fruit set rate of each rice plant is 6.72 seconds, which is much shorter than the manual phenotyping time.
[0066] S4: Input the segmented rice ear image into the ear-grain recognition model to determine the state of the rice ear and grain, identify the rice ear and grain in the classified and labeled rice ear image to obtain a recognition result, and obtain the rice ear and grain related phenotype based on the recognition result.
[0067] Exemplarily, the obtained rice panicle-related phenotypes may be, for example, color characteristics, texture characteristics, rice panicle-grain aspect ratio, and morphological characteristics.
[0068] The above is a rice spike and grain phenotype extraction method provided in one or more embodiments of this specification. Based on the same idea, this specification also provides a corresponding rice spike and grain phenotype extraction system, including:
[0069] An acquisition module is used to acquire the rice ear image of the area to be detected;
[0070] A labeling module is used to segment the rice ear image using multiple instance segmentation models, select the YOLOv8 model with the highest segmentation accuracy as the rice ear image segmentation model, input the rice ear image into the rice ear image segmentation model for segmentation, classify and label the segmented rice ear image according to half-grain, full-grain and full-grain, and determine the classification labeling data;
[0071] The recognition module is used to perform feature recognition on the classified rice ear images using multiple convolutional neural networks based on the classified labeled data, and to compare and analyze the recognition results of each target detection network, and to use the YOLOX network with the highest recognition accuracy as the ear-grain recognition model;
[0072] The determination module is used to input the segmented rice ear image into the ear-grain recognition model, determine the state of the rice ear and grain, identify the rice ear and grain in the classified and labeled rice ear image, obtain the recognition result, and obtain the rice ear and grain related phenotype according to the recognition result.
[0073] For example, Vue can be used as the front-end framework of the system for building the user interface and user interaction, and Django can be used as the back-end framework for processing business logic, database interaction, and server-side rendering; Figure 6 As shown in the figure, it is the functional framework of the system, including the user management module, phenotypic analysis module and data visualization module. Before using other functional modules of the system, users need to complete login or registration. During the registration process, users need to enter a username and password in the specified format. The system will then store this data in the database to complete the registration. The registration and login interface is as follows: Figure 7 and Figure 8 As shown; Fig. 9 This is the rice ear phenotype analysis module page, where users upload rice ear images and select the required phenotype options.
[0074] For example, after entering the phenotypic analysis page, two options are provided: single sheet analysis and multiple sheet analysis. Fig.10 As shown, the operation of uploading pictures is as follows Fig.11 As shown in the figure, when the image is successfully displayed, the user can check the desired phenotype in the checkbox. In the aspect ratio, full and empty are set. You can choose either one or both. After selecting, click Submit. After the progress bar ends, click Data Analysis in the navigation bar to jump to the page. The front end displays the selected phenotype value, indicating that the program functions normally. Fig.12 Shown is the front-end leaflet phenotype results display, Fig.13 It is the result of multiple phenotypes on the front end. The phenotype results of this system provide data export function. Users can select the phenotypes to be exported according to the ID, or they can choose to export all. Fig.14 It is a single analysis data export chart. Fig.15These are multiple analysis data export images; select the texture difference exture_dissimilarity option to visualize all the data. The visualization results of a single data are as follows: Fig.16 As shown in Figure 2, multiple data visualization results are shown in Figure 2. Fig.17 shown.
[0075] For the specific limitations of the rice panicle phenotype extraction system, please refer to the above-mentioned limitations on the rice panicle phenotype extraction method, which will not be repeated here. Each module in the above-mentioned rice panicle phenotype extraction system can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0076] The present invention also provides a computer-readable storage medium, which stores a computer program. The computer program can be used to execute the rice spikelet phenotype extraction method provided above.
[0077] The present invention also provides Fig.18 The structural diagram of the computer device shown in FIG. Fig.18 As shown, at the hardware level, the computer device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include hardware required for other services. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to implement the rice spikelet phenotype extraction method provided in the above embodiment.
[0078] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of the present invention.
Claims
1. A method for extracting rice spikelet phenotypes, characterized in that: include: Acquire the rice ear image of the area to be detected; Multiple instance segmentation models are used to segment the rice ear images. The YOLOv8 model with the highest segmentation accuracy is selected as the rice ear image segmentation model. The rice ear images are input into the rice ear image segmentation model for segmentation. The segmented rice ear images are classified and labeled according to half-grain, full-grain and full-grain, and the classified labeling data is determined. According to the classified labeled data, multiple convolutional neural networks are used to identify the features of the classified rice ear images. Based on the comparative analysis of the recognition results of each target detection network, the YOLOX network with the highest recognition accuracy is used as the ear-grain recognition model. The segmented rice ear image is input into the ear-grain recognition model to determine the state of the rice ear and grain, and the rice ear and grain in the classified and labeled rice ear image are recognized to obtain the recognition result, and the rice ear and grain related phenotype is obtained according to the recognition result.
2. The method for extracting rice spikelet phenotypes according to claim 1, characterized in that: The segmented rice ear images are classified and labeled according to half-grain, full-grain and full-grain deflated, specifically including: The labeling tool Labelme was used to classify and label the segmented rice ear images into half-grain, full-grain and full-grain empty grains.
3. The method for extracting rice spike and grain phenotypes according to claim 1, characterized in that: After determining the state of the rice ears and grains, the rice seed setting rate is obtained using the following formula: Among them, RSSR a is the rice fruit setting rate, NF is the number of whole grains per ear, NE is the number of whole and empty grains per ear, NH is the number of half grains per ear, and PH is the proportion of whole grains per ear.
4. The method for extracting rice spikelet phenotypes according to claim 1, characterized in that: The rice panicle-grain related phenotypes specifically include: Color characteristics, texture characteristics, rice panicle aspect ratio and morphological characteristics.
5. The method for extracting rice spikelet phenotypes according to claim 1, characterized in that: The parameters of the YOLOV8 model are that Epoch is set to 300 and batch size is set to 4.
6. The method for extracting rice spikelet phenotypes according to claim 1, characterized in that: The method further comprises: Before using multiple convolutional neural networks to perform feature recognition on the classified rice ear images, data enhancement processing is performed on the classified and labeled data.
7. A rice spike phenotype extraction system, characterized in that: include: An acquisition module is used to acquire the rice ear image of the area to be detected; A labeling module is used to segment the rice ear image using multiple instance segmentation models, select the YOLOv8 model with the highest segmentation accuracy as the rice ear image segmentation model, input the rice ear image into the rice ear image segmentation model for segmentation, classify and label the segmented rice ear image according to half-grain, full-grain and full-grain, and determine the classification labeling data; The recognition module is used to perform feature recognition on the classified rice ear images using multiple convolutional neural networks based on the classified labeled data, and to compare and analyze the recognition results of each target detection network, and to use the YOLOX network with the highest recognition accuracy as the ear-grain recognition model; The determination module is used to input the segmented rice ear image into the ear-grain recognition model, determine the state of the rice ear and grain, identify the rice ear and grain in the classified and labeled rice ear image, obtain the recognition result, and obtain the rice ear and grain related phenotype according to the recognition result.
8. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the rice panicle and grain phenotype extraction method according to any one of claims 1 to 6 is implemented.
9. A computer device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the rice panicle and grain phenotype extraction method according to any one of claims 1 to 6 is realized.
Citation Information
Cited By
Rice phenotype extraction method integrating deep learning and expert system
CN121366191A