A bullfrog biological behavior recognition optimization system based on deep learning
Through the deep learning-based bullfrog biological behavior recognition optimization system, the biological behavior of bullfrogs is identified and optimized, and the problem of difficult to track and predict bullfrog movement in the prior art is solved, and the accuracy of activity and density monitoring is improved.
Patent Information
- Application Number
- CN202411716098.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-27
- Publication Date
- 2025-05-13
- Estimated Expiration
- 2044-11-27
AI Technical Summary
The prior art is difficult to effectively track and predict a single bullfrog, resulting in an inflated activity monitoring and lack of means of optimizing perturbation.
The bullfrog biological behavior recognition optimization system based on deep learning is adopted. Through the image acquisition module, image recognition module, behavior monitoring module, data optimization module and density evaluation module, the biological behavior of bullfrog is identified and optimized, its theoretical activity coefficient is obtained and optimized, and the movement trajectory and density are predicted.
It improves the accuracy of bullfrog activity and density monitoring, can effectively correct the impact of other bullfrog movements on current bullfrogs, and improves the accuracy of mobile trend prediction.
Smart Images
Figure CN119649454B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of biometric identification technology, and in particular to a bullfrog biological behavior identification optimization system based on deep learning. Background Art
[0002] Using deep learning technology to identify and optimize the biological behavior of bullfrogs is a comprehensive system that integrates advanced artificial intelligence technology and biological knowledge. It aims to achieve accurate and efficient identification and optimization of bullfrog biological behavior. By training and analyzing a large amount of bullfrog behavior data, it can realize automatic identification of bullfrog biological behavior, and can monitor or predict the activity and density of bullfrogs based on the identification results, so as to improve the accuracy of identification and assist bullfrog breeding.
[0003] The behavioral recognition of bullfrogs is mostly based on image data, and the prior art often only recognizes the common features of bullfrogs, but fails to effectively track a single bullfrog, and thus cannot predict its subsequent movement. Since the movement of different bullfrogs often disturbs the movement of other bullfrogs, the monitored activity is falsely higher than it should be, and the prior art lacks means to optimize this disturbance. In view of the shortcomings of the prior art, the present invention provides a bullfrog biological behavior recognition optimization system based on deep learning. Summary of the invention
[0004] The purpose of the present invention is to provide a bullfrog biological behavior recognition optimization system based on deep learning.
[0005] The purpose of the present invention can be achieved by the following technical solution: A bullfrog biological behavior recognition optimization system based on deep learning, comprising the following modules:
[0006] An image acquisition module is used to acquire image data of bullfrog and perform frame decomposition processing on the image data to obtain image data frames;
[0007] An image recognition module is used to extract the first-class feature information and the second-class feature information from the image data frame, and train the image recognition model using the first-class feature information and the second-class feature information;
[0008] The behavior monitoring module is used to set the monitoring period, construct the movement trajectories of different bullfrogs in the same monitoring period, obtain the theoretical activity coefficient of the bullfrog according to the movement trajectory, determine whether the movement trajectories of different bullfrogs are related and obtain the correlation coefficient;
[0009] A data optimization module is used to optimize the theoretical activity coefficient of the bullfrog according to the correlation coefficient to obtain the actual activity coefficient, and input the actual activity coefficient into a preset behavior prediction model to obtain the predicted movement trajectory of the bullfrog;
[0010] The density assessment module is used to divide the activity areas and obtain the predicted bullfrog density in different activity areas based on the predicted movement trajectory.
[0011] Furthermore, the process of collecting the image data of the bullfrog and performing frame decomposition processing on the image data to obtain the image data frame includes:
[0012] The image data refers to a monitoring image that can reflect the appearance characteristics of a bullfrog. The image data is preprocessed, and the preprocessed image data is subjected to frame decomposition processing. The frame decomposition processing refers to dividing the image data into frame segments of a fixed time length, and using the frame segments as image data frames.
[0013] Furthermore, the process of extracting the first-class feature information and the second-class feature information from the image data frame and training the image recognition model using the first-class feature information and the second-class feature information includes:
[0014] The first type of characteristic information refers to the shape features common to all bullfrogs, and the first type of characteristic information of all bullfrogs in the image data frame is extracted using edge detection and contour extraction methods;
[0015] The second type of feature information refers to the unique texture features of each bullfrog, and the second type of feature information of each bullfrog is extracted using the gray level co-occurrence matrix and the local binary pattern method;
[0016] The YOLO algorithm is used as the initial image recognition model, and the initial image recognition model is trained and evaluated using the first-class feature information and the second-class feature information to obtain the image recognition model for bullfrog.
[0017] Furthermore, the process of setting a monitoring period and constructing the movement trajectories of different bullfrogs in the same monitoring period includes:
[0018] The image data frames belonging to the same monitoring period are included in the same image data frame set, and different bullfrogs in the image data frame set are respectively identified and tracked by the image recognition model;
[0019] Each image data frame in the image data frame set is numbered in chronological order, the local area occupied by a single bullfrog in each image data frame is obtained, the local area of the bullfrog in each image data frame is superimposed, and the superimposed area is used as the movement trajectory of the bullfrog in the monitoring period.
[0020] Furthermore, the process of obtaining the theoretical activity coefficient of the bullfrog according to the movement trajectory, determining whether the movement trajectories of different bullfrogs are related and obtaining the correlation coefficient includes:
[0021] The difference in the area of the local regions of two image data frames with adjacent numbers is taken as the activity amplitude of the bullfrog between the two, and all the activity amplitudes S of the bullfrog in a single image data frame set are obtained.i , i = 1, 2, ..., n, n is the number of active amplitudes;
[0022] Obtain the theoretical activity coefficient H of bullfrogs in the corresponding monitoring period; T is the length of the monitoring period;
[0023]
[0024] Compare the movement trajectories of any two bullfrogs. If the local areas of the two bullfrogs in the same image data frame have an intersection, the two bullfrogs are judged to be related in the image data frame, and the intersection area of the local areas is used as the correlation amplitude of the two bullfrogs in the image data frame.
[0025] Get all the correlation amplitudes Q between any two bullfrogs in a single image data frame set j , j = 1, 2, ..., m, where m is the number of correlation amplitudes, and the correlation coefficient G between the two within the monitoring period is obtained;
[0026]
[0027] Furthermore, the process of optimizing the theoretical activity coefficient of bullfrog according to the correlation coefficient to obtain the actual activity coefficient includes:
[0028] Obtain the correlation coefficients G of a single bullfrog in a single monitoring period k , k = 1, 2, ..., b, b is the number of correlation coefficients, and the theoretical activity coefficient H of the bullfrog corresponding to each correlation coefficient in the monitoring period is obtained k ;
[0029] According to the correlation coefficients and their corresponding theoretical activity coefficients, the theoretical activity coefficient H of the bullfrog 理 Optimize and obtain the actual activity coefficient H 实 ;
[0030]
[0031] Furthermore, the actual activity coefficient is input into a preset behavior prediction model, and the process of obtaining the predicted movement trajectory of the bullfrog includes:
[0032] Generate a behavior prediction set according to the theoretical activity coefficients of different bullfrogs and the activity amplitudes under monitoring time, wherein the monitoring time refers to the time interval between the corresponding time of each activity amplitude and the start time of the monitoring cycle, and divide the behavior prediction set into a training set and a test set;
[0033] Construct a convolutional neural network, use the theoretical activity coefficient and monitoring duration in the training set as input data of the convolutional neural network, use the corresponding activity amplitude in the training set as output data of the convolutional neural network, and train the convolutional neural network to obtain an initial convolutional neural network;
[0034] The initial convolutional neural network is verified using the test set, and the initial convolutional neural network with a value less than or equal to the preset test error threshold is output as the behavior prediction model. The actual activity coefficient and predicted duration of the bullfrog are input into the behavior prediction model to obtain the corresponding predicted activity amplitude.
[0035] The predicted duration refers to the time interval between the corresponding moment of the expected prediction and the current moment. The moving direction of the bullfrog is obtained in real time through the image recognition model, and the predicted activity amplitude is continuously superimposed on the moving direction to obtain the predicted movement trajectory of the bullfrog at the predicted duration.
[0036] Furthermore, the process of dividing the activity areas and obtaining the predicted bullfrog density in different activity areas according to the predicted movement trajectory includes:
[0037] A plurality of activity areas are divided in the image data frame, the number of bullfrogs in each activity area is monitored in real time, and the ratio of the number of bullfrogs to the area of the activity area is used as the bullfrog density of the activity area;
[0038] The predicted activity trajectory of each bullfrog at the predicted duration is displayed on the current image data frame, the predicted position of each bullfrog at the predicted duration is obtained, the predicted number of bullfrogs in each activity area at the predicted duration is obtained according to the predicted position, and the corresponding predicted bullfrog density is obtained.
[0039] Compared with the prior art, the present invention has the following beneficial effects:
[0040] 1. The present invention can not only identify the shape features of bullfrogs, but also identify the texture features of different bullfrogs by training an image recognition model. By decomposing the image data frame of the bullfrog into separate image data frames, the local area of the bullfrog in each image data frame is obtained, and the movement trajectory of the bullfrog can be constructed according to the local areas of the bullfrog in different image data frames, and the corresponding theoretical activity coefficient can be obtained, so as to improve the accuracy of monitoring the activity of the bullfrog;
[0041] 2. By comparing the movement trajectories of any two bullfrogs, it is possible to determine whether there is a correlation in the movement trajectories of the bullfrogs, and then obtain the correlation coefficient of any two bullfrogs. The theoretical activity coefficient can be optimized using the correlation coefficient to obtain the actual activity coefficient, which can effectively correct the impact of the movement of other bullfrogs on the current bullfrog. By constructing a behavior prediction model, the predicted movement trajectory at different prediction time lengths can be obtained according to the actual activity coefficient, which is conducive to predicting the movement trend of the bullfrog and can improve the accuracy of the bullfrog density prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION
[0043] like Figure 1 As shown, a bullfrog biological behavior recognition optimization system based on deep learning includes the following modules:
[0044] An image acquisition module is used to acquire image data of bullfrog and perform frame decomposition processing on the image data to obtain image data frames;
[0045] An image recognition module is used to extract the first-class feature information and the second-class feature information from the image data frame, and train the image recognition model using the first-class feature information and the second-class feature information;
[0046] The behavior monitoring module is used to set the monitoring period, construct the movement trajectories of different bullfrogs in the same monitoring period, obtain the theoretical activity coefficient of the bullfrog according to the movement trajectory, determine whether the movement trajectories of different bullfrogs are related and obtain the correlation coefficient;
[0047] A data optimization module is used to optimize the theoretical activity coefficient of the bullfrog according to the correlation coefficient to obtain the actual activity coefficient, and input the actual activity coefficient into a preset behavior prediction model to obtain the predicted movement trajectory of the bullfrog;
[0048] The density assessment module is used to divide the activity areas and obtain the predicted bullfrog density in different activity areas based on the predicted movement trajectory.
[0049] It should be further explained that, in the specific implementation process, the process of collecting the image data of the bullfrog and performing frame decomposition processing on the image data to obtain the image data frame includes:
[0050] Setting a monitoring unit, collecting image data of the bullfrog through the monitoring unit, wherein the image data refers to a monitoring image that can reflect the appearance characteristics of the bullfrog;
[0051] Setting a processing unit, through which the collected image data is preprocessed, the preprocessing includes grayscale, geometric transformation, image enhancement, etc., which is used to improve the clarity of the image data;
[0052] A decomposition unit is set up, and frame decomposition processing is performed on the pre-processed image data by the decomposition unit. The frame decomposition processing refers to dividing the continuous image data into a series of separate frame segments with a fixed time length, and marking the obtained frame segments as image data frames.
[0053] It should be further explained that, in a specific implementation process, the process of extracting the first-class feature information and the second-class feature information from the image data frame and using the first-class feature information and the second-class feature information to train the image recognition model includes:
[0054] Taking any image data frame as an example, there are an indefinite number of bullfrogs in each image data frame at the same time, and the type of feature information refers to the shape features shared by all bullfrogs, which is used to reflect the overall shape and contour shared by different bullfrogs, and the type of feature information of all bullfrogs in the image data frame is extracted by edge detection and contour extraction methods;
[0055] The second type of feature information refers to the unique texture features of each bullfrog, which is used to reflect the unique skin patterns and roughness of different bullfrogs. The second type of feature information of each bullfrog is extracted using methods such as gray level co-occurrence matrix and local binary pattern;
[0056] The present invention adopts the YOLO algorithm as the initial image recognition model. The YOLO algorithm (You Only Look Once) is a target detection algorithm based on deep learning. The initial image recognition model is trained and evaluated using the extracted first-class feature information and second-class feature information to obtain an image recognition model for bullfrogs.
[0057] The image recognition model can identify all bullfrogs in subsequent image data frames, can also track each identified bullfrog, and continuously update the image recognition model using the first-category feature information and the second-category feature information obtained subsequently.
[0058] It should be further explained that, in the specific implementation process, the process of setting the monitoring period and constructing the movement trajectories of different bullfrogs in the same monitoring period includes:
[0059] Setting a monitoring period, including image data frames belonging to the same monitoring period into the same image data frame set, and binding each image data frame set to its corresponding monitoring period, and respectively identifying and tracking different bullfrogs in the image data frame set by using the image recognition model;
[0060] Taking any bullfrog as an example, each image data frame in the image data frame set is numbered according to its time sequence, and the local area occupied by the bullfrog in each image data frame is obtained, and the local area is used to reflect the position of the bullfrog in different image data frames;
[0061] The local areas of the bullfrog in each image data frame are superimposed to obtain the corresponding superimposed area, and the obtained superimposed area is used as the movement trajectory of the bullfrog in the monitoring period. The same method is adopted to obtain the movement trajectories of other bullfrogs in the monitoring period.
[0062] It should be further explained that, in the specific implementation process, the process of obtaining the theoretical activity coefficient of the bullfrog according to the movement trajectory, determining whether the movement trajectories of different bullfrogs are related and obtaining the correlation coefficient includes:
[0063] Taking the movement trajectory of any bullfrog as an example, the movement trajectory is formed by superimposing local areas in different image data frames, and the difference in area of the local areas of two image data frames with adjacent numbers is taken as the activity amplitude of the bullfrog between the two;
[0064] Get all the active amplitudes of the bullfrog in the image data frame set, denoted as S i , i = 1, 2, ..., n, n is the number of active amplitudes, and the theoretical activity coefficient of the bullfrog in the monitoring period is obtained, which is recorded as H;
[0065]
[0066] Among them, T represents the length of the monitoring period, and the same method is used to obtain the theoretical activity coefficients of different bullfrogs in the same monitoring period;
[0067] Taking the movement trajectories of any two bullfrogs as an example, if the local areas of the two bullfrogs in the same image data frame have an intersection, then the two bullfrogs are judged to be related in the image data frame, and the intersection area of the local areas is used as the correlation amplitude of the two bullfrogs in the image data frame; if the intersection area does not exist, then the bullfrogs are judged to be unrelated.
[0068] Get all the correlation amplitudes of the two bullfrogs in the image data frame set, recorded as Q j , j = 1, 2, ..., m, where m is the number of correlation amplitudes, and the correlation coefficient between the two in the monitoring period is obtained, which is recorded as G;
[0069]
[0070] Similarly, T represents the length of the monitoring period, and the same method is used to obtain the correlation coefficient between any two bullfrogs in the same monitoring period.
[0071] It should be further explained that, in the specific implementation process, the process of optimizing the theoretical activity coefficient of bullfrog according to the correlation coefficient to obtain the actual activity coefficient includes:
[0072] Taking any bullfrog in any monitoring period as an example, obtain the correlation coefficients of the bullfrog in the monitoring period, denoted as G k , k = 1, 2, ..., b, b is the number of correlation coefficients, and the theoretical activity coefficient of another bullfrog corresponding to each correlation coefficient in the monitoring period is obtained, which is recorded as H k , where G k With H k Corresponding respectively;
[0073] According to the above correlation coefficient and its corresponding theoretical activity coefficient, the theoretical activity coefficient H of the bullfrog 理 Optimize and obtain the corresponding actual activity coefficient, recorded as H 实 ;
[0074]
[0075] The same method was used to optimize the theoretical activity coefficient of each bullfrog in different monitoring periods, and to obtain its corresponding actual activity coefficient.
[0076] It should be further explained that, in the specific implementation process, the actual activity coefficient is input into the preset behavior prediction model, and the process of obtaining the predicted movement trajectory of the bullfrog includes:
[0077] The preset process of the behavior prediction model is as follows:
[0078] Generate a behavior prediction set according to the theoretical activity coefficients of different bullfrogs and the activity amplitude under the monitoring time, and divide the obtained behavior prediction set into a training set and a test set, wherein the monitoring time refers to the time interval between the corresponding moment of each activity amplitude and the start moment of the monitoring cycle;
[0079] Construct a convolutional neural network, use the theoretical activity coefficient and monitoring duration in the training set as input data of the convolutional neural network, use the corresponding activity amplitude in the training set as output data of the convolutional neural network, and train the convolutional neural network to obtain the corresponding initial convolutional neural network;
[0080] The initial convolutional neural network is model verified using the test set, and the initial convolutional neural network with a test error threshold value less than or equal to a preset value is output as the corresponding behavior prediction model;
[0081] The actual activity coefficient and predicted duration of the bullfrog are input into the behavior prediction model to obtain the corresponding predicted activity amplitude. The predicted duration refers to the time interval between the corresponding moment of the expected prediction and the current moment. The moving direction of the bullfrog, that is, the head direction of the bullfrog, is obtained in real time through the image recognition model. The predicted activity amplitude is continuously superimposed on the obtained moving direction to obtain the predicted movement trajectory of the bullfrog at the predicted duration.
[0082] It should be further explained that, in the specific implementation process, the process of dividing the activity area and obtaining the predicted bullfrog density in different activity areas according to the predicted movement trajectory includes:
[0083] Taking any image data frame as an example, a number of sub-areas, namely, activity areas, of equal area are divided in the image data frame, the number of bullfrogs in each activity area is monitored in real time, and the ratio of the number of bullfrogs to the area of the activity area is taken as the bullfrog density of the activity area;
[0084] The predicted activity trajectory of each bullfrog at the predicted duration is displayed on the current image data frame, the predicted position of each bullfrog at the predicted duration is obtained, the predicted number of bullfrogs in each activity area at the predicted duration is obtained according to the predicted position, and the corresponding predicted bullfrog density is obtained.
[0085] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.
Claims
1. A bullfrog biological behavior recognition optimization system based on deep learning, characterized in that: Includes the following modules: An image acquisition module is used to acquire image data of bullfrog and perform frame decomposition processing on the image data to obtain image data frames; An image recognition module is used to extract the first-class feature information and the second-class feature information from the image data frame, and train the image recognition model using the first-class feature information and the second-class feature information; The behavior monitoring module is used to set the monitoring period, construct the movement trajectories of different bullfrogs in the same monitoring period, obtain the theoretical activity coefficient of the bullfrog according to the movement trajectory, determine whether the movement trajectories of different bullfrogs are related and obtain the correlation coefficient; A data optimization module is used to optimize the theoretical activity coefficient of the bullfrog according to the correlation coefficient to obtain the actual activity coefficient, and input the actual activity coefficient into a preset behavior prediction model to obtain the predicted movement trajectory of the bullfrog; Density assessment module, used to divide activity areas and obtain predicted bullfrog densities in different activity areas based on predicted movement trajectories; The process of collecting bullfrog image data and obtaining image data frames includes: The image data refers to a monitoring image that can reflect the appearance characteristics of the bullfrog, the image data is preprocessed, and the preprocessed image data is subjected to frame decomposition processing, wherein the frame decomposition processing refers to dividing the image data into frame segments of a fixed time length, and using the frame segments as image data frames; The process of training an image recognition model using first-class feature information and second-class feature information includes: The first type of characteristic information refers to the shape features common to all bullfrogs, and the first type of characteristic information of all bullfrogs in the image data frame is extracted using edge detection and contour extraction methods; The second type of feature information refers to the unique texture features of each bullfrog, and the second type of feature information of each bullfrog is extracted using the gray level co-occurrence matrix and the local binary pattern method; The YOLO algorithm is used as the initial image recognition model, and the initial image recognition model is trained and evaluated using the first-class feature information and the second-class feature information to obtain the image recognition model for bullfrogs; The process of constructing the movement trajectories of different bullfrogs within the same monitoring period includes: The image data frames belonging to the same monitoring period are included in the same image data frame set, and different bullfrogs in the image data frame set are respectively identified and tracked by the image recognition model; Each image data frame in the image data frame set is numbered in chronological order, the local area occupied by a single bullfrog in each image data frame is obtained, the local area of the bullfrog in each image data frame is superimposed, and the superimposed area is used as the movement trajectory of the bullfrog in the monitoring period.
2. A bullfrog biological behavior recognition optimization system based on deep learning according to claim 1, characterized in that: The process of obtaining the theoretical activity coefficient and correlation coefficient of bullfrog includes: The difference in the area of the local regions of two image data frames with adjacent numbers is taken as the activity amplitude of the bullfrog between the two, and all the activity amplitudes S of the bullfrog in a single image data frame set are obtained. i , i = 1, 2, ..., n, n is the number of active amplitudes; Obtain the theoretical activity coefficient H of bullfrogs in the corresponding monitoring period, where T is the length of the monitoring period; Compare the movement trajectories of any two bullfrogs. If the local areas of the two bullfrogs in the same image data frame have an intersection, the two bullfrogs are judged to be related in the image data frame, and the intersection area of the local areas is used as the correlation amplitude of the two bullfrogs in the image data frame. Get all the correlation amplitudes Q between any two bullfrogs in a single image data frame set j , j = 1, 2, ..., m, where m is the number of correlation amplitudes, and the correlation coefficient G between the two within the monitoring period is obtained; 3. A bullfrog biological behavior recognition optimization system based on deep learning according to claim 2, characterized in that: The process of obtaining the actual activity coefficient includes: Obtain the correlation coefficients G of a single bullfrog in a single monitoring period k , k = 1, 2, ..., b, b is the number of correlation coefficients, and the theoretical activity coefficient H of the bullfrog corresponding to each correlation coefficient in the monitoring period is obtained k ; According to the correlation coefficients and their corresponding theoretical activity coefficients, the theoretical activity coefficient H of the bullfrog 理 Optimize and obtain the actual activity coefficient H 实 ; 4. The bullfrog biological behavior recognition optimization system based on deep learning according to claim 3 is characterized in that: The process of obtaining the predicted movement trajectory of the bullfrog includes: Generate a behavior prediction set according to the theoretical activity coefficients of different bullfrogs and the activity amplitudes under monitoring time, wherein the monitoring time refers to the time interval between the corresponding time of each activity amplitude and the start time of the monitoring cycle, and divide the behavior prediction set into a training set and a test set; Construct a convolutional neural network, use the theoretical activity coefficient and monitoring duration in the training set as input data of the convolutional neural network, use the corresponding activity amplitude in the training set as output data of the convolutional neural network, and train the convolutional neural network to obtain an initial convolutional neural network; The initial convolutional neural network is verified using the test set, and the initial convolutional neural network with a value less than or equal to the preset test error threshold is output as the behavior prediction model. The actual activity coefficient and predicted duration of the bullfrog are input into the behavior prediction model to obtain the corresponding predicted activity amplitude. The predicted duration refers to the time interval between the corresponding moment of the expected prediction and the current moment. The moving direction of the bullfrog is obtained in real time through the image recognition model, and the predicted activity amplitude is continuously superimposed on the moving direction to obtain the predicted movement trajectory of the bullfrog at the predicted duration.
5. A bullfrog biological behavior recognition optimization system based on deep learning according to claim 4, characterized in that: The process of obtaining the predicted bullfrog density in different activity areas based on the predicted movement trajectory includes: A plurality of activity areas are divided in the image data frame, the number of bullfrogs in each activity area is monitored in real time, and the ratio of the number of bullfrogs to the area of the activity area is used as the bullfrog density of the activity area; The predicted activity trajectory of each bullfrog at the predicted duration is displayed on the current image data frame, the predicted position of each bullfrog at the predicted duration is obtained, the predicted number of bullfrogs in each activity area at the predicted duration is obtained according to the predicted position, and the corresponding predicted bullfrog density is obtained.
Citation Information
Patent Citations
Area determination method and device, storage medium and processor
CN109919966A
Quantitative simulation method for motion behaviors of frog animals
CN114494337A