Method and system for intelligently identifying rice growth vigor based on video monitoring

Abstract

The invention discloses a method and system for intelligently identifying rice growth based on video monitoring, and belongs to the technical field of growth period prediction.The method comprises the steps that image data of rice is obtained through video monitoring equipment, and the shooting height value of the rice is extracted; establishing a height prediction model to predict and obtain a predicted height data set; evaluating the height dispersion degree of the rice by using the prediction height data set and the shooting height value to obtain an evaluation value; if the evaluation value is greater than a first threshold value, extracting and outputting current growth period data and a shooting height value by adopting an image recognition method; if the evaluation value is smaller than or equal to a first threshold value, generating and outputting current growth period data and predicted height by adopting a growth simulation method; and generating a rice growth report by using the output data. Video monitoring is combined with an intelligent algorithm, the rice height dispersion degree is evaluated through the prediction model, the growth period data is acquired by selecting an image recognition or growth simulation method, and the efficiency and accuracy of rice growth vigor recognition can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of growth period prediction, and in particular to a method and system for intelligently identifying rice growth based on video monitoring. Background Art

[0002] At present, the monitoring of rice growth mainly relies on manual field observation and traditional agricultural sensor monitoring methods. Although manual field observation can obtain relatively accurate rice growth information, this method is time-consuming and labor-intensive, and is limited by weather, terrain and other conditions, making it difficult to achieve large-scale, high-efficiency monitoring. Although traditional agricultural sensor monitoring methods can obtain real-time data on the rice growth environment, such as temperature, humidity, and light, these data cannot directly reflect the growth of rice, and the deployment and maintenance costs of sensors are relatively high.

[0003] With the rapid development of computer vision and artificial intelligence technology, intelligent recognition technology based on image and video monitoring has gradually shown great potential in rice growth monitoring. However, most of the existing rice growth recognition methods have problems such as low recognition accuracy and low recognition efficiency. On the one hand, due to the complexity of the rice growth environment, such as lighting changes, the presence of obstructions and other factors, it is difficult to extract rice information from video images, which in turn affects the accuracy of recognition. On the other hand, traditional recognition methods often require frame-by-frame processing of video images, which has a large amount of calculation and slow processing speed, and it is difficult to meet the needs of real-time monitoring. Therefore, how to develop a method and system that can efficiently and accurately identify rice growth has become a problem that needs to be solved in the current field of agricultural informatization and intelligence. Summary of the invention

[0004] To solve the above problems, the present invention provides a method and system for intelligently identifying the growth of rice based on video monitoring. The method combines video monitoring with an intelligent algorithm, evaluates the degree of rice height discreteness through a prediction model, and selects image recognition or growth simulation method to obtain growth period data, which can improve the efficiency and accuracy of rice growth identification.

[0005] The above objectives can be achieved through the following solutions: A method for intelligently identifying rice growth based on video monitoring, comprising: obtaining image data of rice using video monitoring equipment, and extracting shooting height values ​​of rice using the image data; establishing a height prediction model for predicting rice height, and predicting rice height data at different time points using the height prediction model to obtain a predicted height data set; evaluating the height discreteness of rice using the predicted height data set and the shooting height value to obtain an evaluation value; judging whether the evaluation value is greater than a preset first threshold; if so, extracting and outputting current growth period data and the shooting height value using an image recognition method; if not, generating and outputting current growth period data and predicted height using a growth simulation method; and generating a rice growth report using the output data.

[0006] Optionally, the method of establishing a height prediction model for predicting the height of rice and using the height prediction model to predict the rice height data at different time points to obtain a predicted height data set includes: collecting historical growing days of rice, historical environmental factor data, and historical growth heights corresponding to historical growing days to construct a historical data set; using the growing days and environmental factor data as input and the growth height as output, and using the historical data set to establish and train a neural network model to obtain a height prediction model; collecting the growing days and environmental factor data of rice by day to obtain an input data set; inputting the input data set into the height prediction model to obtain predicted values ​​of rice height corresponding to different growing days; and integrating the predicted values ​​of rice height corresponding to different growing days to obtain a predicted height data set.

[0007] Optionally, the use of the predicted height data set and the photographed height value to evaluate the height dispersion of rice to obtain the evaluation value comprises: using the predicted height data set to calculate the mean of the predicted rice height values; using the predicted height data set and the mean to calculate the standard deviation of the predicted rice height values; establishing an evaluation function for evaluating the height dispersion of rice, and inputting the mean, the standard deviation and the photographed height value into the evaluation function to calculate the evaluation value, and for the evaluation value ,have , In the formula, is the shooting height value, is the mean of the predicted values ​​of rice height, is the standard deviation of the predicted value of rice height.

[0008] Optionally, the method of using an image recognition method to extract and output the current growth period data and the shooting height value includes: acquiring image data of rice at different growth periods from a preset rice growth period image database to establish an image data set; extracting feature data from the image data of rice at different growth periods, and constructing a first training set using the feature data and the growth period corresponding to the feature data; using the feature data as input and the growth period as output, and using the first training set to establish and train a neural network model to obtain a growth period recognition model.

[0009] Optionally, the method of using image recognition to extract and output the current growth period data and the shooting height value also includes: obtaining current image data of rice using video monitoring equipment; performing feature extraction on the current image data of rice to obtain current feature data; inputting the current feature data into the growth period recognition model to obtain current growth period information; and outputting the current growth period information and the shooting height value.

[0010] Optionally, the use of the output data to generate a rice growth report includes: inputting actual growth period information and an actual height value; calculating a first difference between the actual height value and the photographed height value; when the first difference is greater than a preset second threshold, adjusting the video monitoring equipment; comparing and analyzing the actual growth period information with the output current growth period information; when the actual growth period information is inconsistent with the output current growth period information, optimizing the parameters of the growth period recognition model.

[0011] Optionally, the use of the growth simulation method to generate and output the current growth period data and predicted height includes: obtaining the historical growth days, historical height data, historical environmental factor data, and historical growth period information of rice to construct a second training set; using the growth days, height data, and environmental factor data as input, and the growth period information of the next growth days as output, and using the second training set to establish and train a model to obtain a growth simulation model; obtaining the current environmental factor data of rice, the current growth days, and the current predicted value of rice height to obtain a current data set; inputting the current data set into the growth simulation model to obtain the growth period information of the next growth days, and inputting it into a preset time series database; obtaining the current growth days, and obtaining the growth period information corresponding to the current growth days from the time series database; outputting the growth period information corresponding to the current growth days and the current predicted value of rice height.

[0012] Optionally, the method of generating a rice growth report using the output data also includes: inputting actual growth period information and an actual height value; calculating a second difference between the actual height value and a current predicted rice height value; determining whether the second difference is greater than a preset third threshold; and if so, optimizing the parameters of the height prediction model.

[0013] Optionally, the determining whether the second difference is greater than a preset third threshold value also includes: if the second difference is less than or equal to the third threshold value, determining whether the actual growth period information is consistent with the growth period information corresponding to the output current growth days; if not, optimizing the growth simulation model and using the optimized growth simulation model to update the time series database.

[0014] Based on the same inventive concept, the present invention also provides a system for intelligently identifying rice growth based on video monitoring, the system comprising: a rice height acquisition module, used to obtain rice image data using video monitoring equipment, and extract the shooting height value of rice using the image data; a rice height prediction module, used to establish a height prediction model for predicting rice height, and use the height prediction model to predict rice height data at different time points to obtain a predicted height data set; a discrete degree evaluation module, used to evaluate the height discreteness of rice using the predicted height data set and the shooting height value to obtain an evaluation value; an evaluation value judgment module, used to judge whether the evaluation value is greater than a preset first threshold; an image recognition module, used to extract and output the current growth period data and the shooting height value using an image recognition method if the evaluation value is greater than the first threshold; a growth simulation module, used to generate and output the current growth period data and the predicted height using a growth simulation method if the evaluation value is less than or equal to the first threshold; a growth report generation module, used to generate a rice growth report using the output data.

[0015] Compared with the prior art, the present invention has the following advantages: 1. The present invention determines the stability of the rice growth state by evaluating the height dispersion of rice. When the stability is good, the growth simulation model is used for prediction, which not only ensures the recognition accuracy but also improves the recognition efficiency. When the stability is poor, the image recognition method can more accurately identify the current growth period. 2. The present invention can predict the growth period information of a certain number of future growth days through the growth simulation model; this prediction method based on historical data does not require real-time image processing and recognition, thereby greatly improving the recognition efficiency; 3. The present invention uses an image recognition method to obtain real-time image data of rice using video monitoring equipment, and uses a pre-trained growth period recognition model to perform feature extraction and growth period recognition, so as to obtain accurate growth period information in a relatively short time; 4. The present invention also provides a method for optimizing parameters such as height prediction models, growth period identification models and growth simulation models; by inputting actual growth period information and actual height values, and comparing and analyzing them with the prediction results, the model parameters can be adjusted and optimized, thereby improving the prediction accuracy and applicability of the model.

[0016] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 The present invention is a flowchart of a method for intelligently identifying rice growth based on video monitoring according to an embodiment of the present invention.

[0019] Figure 2 The present invention is a flowchart of an implementation method of a method for intelligently identifying rice growth based on video monitoring according to an embodiment of the present invention.

[0020] Figure 3 It is an execution flow chart of the image recognition method according to an embodiment of the present invention.

[0021] Figure 4 It is an execution flow chart of the growth simulation method according to an embodiment of the present invention.

[0022] Figure 5 The present invention is a schematic diagram of the structure of a system for intelligently identifying rice growth based on video monitoring according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments 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.

[0024] Reference Figure 1 An embodiment of the present invention proposes a method for intelligently identifying rice growth based on video monitoring, which uses video monitoring combined with intelligent algorithms, evaluates the degree of rice height dispersion through a prediction model, and selects image recognition or growth simulation methods to obtain growth period data, which can improve the efficiency and accuracy of rice growth identification.

[0025] The method of this embodiment specifically includes: Acquiring image data of rice using video monitoring equipment, and extracting a shooting height value of the rice using the image data; Establishing a height prediction model for predicting rice height, and using the height prediction model to predict rice height data at different time points to obtain a predicted height data set; Using the predicted height data set and the photographed height value, evaluating the height dispersion of rice to obtain an evaluation value; Determining whether the evaluation value is greater than a preset first threshold; If so, the image recognition method is used to extract and output the current growth period data and the shooting height value; If not, the growth simulation method is used to generate and output the current growth period data and predicted height; Use the output data to generate a rice growth report.

[0026] Specifically, Figure 2 As shown, video monitoring equipment is used to obtain image data of rice, and the growth of rice is intelligently identified using a height prediction model and a growth period recognition model; first, the image data of rice is obtained through a video monitoring device, and the shooting height value of rice is extracted; then a height prediction model is established, and the historical data set is used for training to predict the rice height data at different time points to obtain a predicted height data set; then the predicted height data set and the shooting height value are used to evaluate the height discreteness of rice, and according to the evaluation value, an image recognition method or a growth simulation method is selected to extract and output the current growth period data and the shooting height value; finally, a rice growth report is generated using the output data; this method can realize real-time and accurate monitoring of rice growth, improve recognition accuracy and efficiency, and provide strong technical support for agricultural production.

[0027] Optionally, the step of establishing a height prediction model for predicting rice height, and using the height prediction model to predict rice height data at different time points to obtain a predicted height data set comprises: Collect historical rice growing days, historical environmental factor data, and historical growth heights corresponding to historical growing days to construct a historical data set; Specifically, historical environmental factor data may include temperature, humidity, light, management measures, etc., wherein management measures include fertilization, irrigation, etc.

[0028] Taking the growth days and environmental factor data as input and the growth height as output, a neural network model is established and trained using the historical data set to obtain a height prediction model; Specifically, the selected neural network model is trained using historical data sets; by continuously adjusting the model parameters (such as weights and biases), the error between the predicted height and the actual height is minimized, thereby obtaining an optimized height prediction model.

[0029] The data of rice growth days and environmental factors are collected by day to obtain the input data set; Inputting the input data set into the height prediction model to obtain predicted values ​​of rice height corresponding to different growing days; The predicted values ​​of rice height corresponding to different growing days are integrated to obtain the predicted height data set.

[0030] Exemplarily, an LSTM model is selected as a height prediction model, with growing days, temperature, humidity, light, fertilization, and irrigation as input features and growth height as the output target. The LSTM model is trained using a historical data set, and the parameters of the model are adjusted through a back-propagation algorithm until the prediction error reaches an acceptable range. The current environmental factor data (temperature, humidity, light, fertilization, irrigation) and the current growing days are collected, and these data are input into the trained LSTM model to generate the current rice height prediction value and store it in the predicted height data set. The predicted height data set stores all rice height prediction values ​​from the past to the present. This data set will be used for subsequent rice growth evaluation and analysis.

[0031] Optionally, the using the predicted height data set and the photographed height value to evaluate the height dispersion of rice to obtain the evaluation value includes: Calculate the mean of the predicted values ​​of rice height using the predicted height data set; Specifically, the mean reflects the average level of predicted rice heights. By calculating the average of all predicted heights, a benchmark value representing the overall height can be obtained.

[0032] Calculate the standard deviation of the predicted value of rice height using the predicted height data set and the mean; Specifically, the standard deviation measures the dispersion of the predicted values ​​of rice height from their mean.

[0033] Establish an evaluation function for evaluating the degree of dispersion of rice height, input the mean, the standard deviation and the shooting height value into the evaluation function, calculate and obtain an evaluation value, and for the evaluation value ,have , In the formula, is the shooting height value, is the mean of the predicted values ​​of rice height, is the standard deviation of the predicted value of rice height.

[0034] Specifically, the evaluation function is used to quantify the relative difference between the shooting height value and the mean predicted height, and takes into account the standard deviation of the predicted height; this evaluation value reflects the degree of deviation of the current shooting height relative to the overall predicted height; when the evaluation value is large, it means that the deviation between the current shooting height and the overall predicted height is large, which may indicate that the growth state of rice is affected by certain external factors, resulting in unstable growth.

[0035] Optionally, the extracting and outputting the current growth period data and the shooting height value by using an image recognition method includes: Acquire image data of rice at different growth stages from a preset rice growth stage image database to establish an image data set; Specifically, image data of rice at different growth stages are obtained from a preset rice growth stage image database. These image data should cover the entire growth cycle of rice, including all stages from sowing to maturity; by collecting a large number of rice images marked with different growth stages, a comprehensive image dataset is constructed.

[0036] Extracting feature data from image data of rice at different growth stages, and constructing a first training set using the feature data and the growth stages corresponding to the feature data; Specifically, feature extraction is a key step in image recognition, which involves extracting information from the image that is useful for growth period recognition. These features can be color, texture, shape, etc.; the training set is composed of feature data and corresponding growth period labels. In this step, the feature data of each image needs to be matched with its corresponding growth period label to form the first training set; the quality of the first training set directly affects the training effect and recognition accuracy of the model.

[0037] Taking the characteristic data as input and the growth period as output, a neural network model is established and trained using the first training set to obtain a growth period recognition model.

[0038] Specifically, the extracted feature data is used as input, and the corresponding growth period is used as output, and a neural network model is established and trained using the first training set. By continuously adjusting the parameters of the model (such as weights and biases), the model can accurately predict the growth period of rice based on the input feature data.

[0039] For example, suppose there is a database of rice growth images, which contains images of rice from 7 days, 14 days, 21 days after sowing... to maturity; 1,000 images are selected from this database, and each image is labeled with the corresponding rice growth stage, such as "14 days after sowing", "tillering stage", etc., to construct a data set containing 1,000 images; use image processing technology, such as OpenCV, to extract features from each image; for example, in the image of "14 days after sowing", feature data representing the color, texture and shape of rice seedlings may be extracted; then, these feature data are compared with the image of "14 days after sowing" The first training set is used to associate the feature data with the growth period label of "14 days" to form a training sample; this process is repeated until all images have been processed, and finally a first training set containing feature data and growth period labels is obtained; a neural network model suitable for image recognition is selected, such as a convolutional neural network (CNN); the feature data and growth period labels in the first training set are used as input and output to train the CNN; during the training process, the back propagation algorithm is used to adjust the parameters of the CNN until the prediction accuracy of the model reaches an acceptable level; for example, after multiple iterations of training, the prediction accuracy of the model for the test set reaches more than 95%.

[0040] Alternatively, if Figure 3 As shown, the method of using image recognition to extract and output the current growth period data and the shooting height value also includes: Use video monitoring equipment to obtain current image data of rice; Specifically, video monitoring equipment (such as cameras) deployed in rice fields can be used to capture rice image data in real time. These image data should contain sufficient information for subsequent feature extraction and growth period identification.

[0041] Extract features from the current image data of rice to obtain current feature data; Specifically, after receiving the current image data of rice, image processing technology (such as OpenCV, etc.) is used to extract features of the image; these features can be color, texture, shape, etc., which are crucial for distinguishing different growth stages.

[0042] Inputting the current characteristic data into the growth period recognition model to obtain the current growth period information; Specifically, the extracted current feature data is input into a pre-trained growth period recognition model, which is trained based on a large amount of rice image data marked with different growth periods, and can accurately predict the growth period of rice based on the input feature data.

[0043] Output the current growth period information and the shooting height value.

[0044] Specifically, in addition to outputting the current growth period information, the shooting height value of the rice is extracted using the image data, and finally the growth period information and the shooting height value are output together for subsequent analysis and processing.

[0045] For example, assume that a high-definition camera is installed in a rice field, and the camera automatically takes an image of the rice at regular intervals (such as every hour); at a certain point in time, the camera captures a current image of the rice and transmits it to the system for processing; OpenCV is used to process the received rice image to extract feature data representing the current growth status of the rice; for example, the color, texture, and overall morphology of the rice leaves are identified in the image, and these features will be used as input for subsequent growth period recognition; assume that a convolutional neural network (CNN) model has been trained as a growth period recognition model using a data set containing 1,000 images, and now the currently extracted feature data is input into the CNN model, and the model outputs the current growth period information after calculation, such as "mid-tillering period"; the shooting height of the rice is calculated to be 120 cm based on the image data, and this height value is output together with the previously obtained growth period information "mid-tillering period", and these data will be used to generate a rice growth report to provide decision support for agricultural production.

[0046] Alternatively, if Figure 3 As shown, the method of generating a rice growth report using the output data includes: Enter the actual growth period information and actual height value; Specifically, it is necessary to input the actual rice growth period information and the actual height value, which can be obtained through manual field observation or other reliable means for comparison and verification with the output data.

[0047] Calculating a first difference between the actual height value and the photographed height value; Specifically, the difference between the actual height value and the photographed height value is calculated to evaluate the accuracy and stability of the video monitoring equipment.

[0048] When the first difference is greater than a preset second threshold, adjusting the video monitoring device; Specifically, if the calculated first difference is greater than a preset second threshold (such as 5 centimeters), it means that the video monitoring device may have errors or be unstable, and the device needs to be adjusted or calibrated.

[0049] Comparing and analyzing the actual growth period information with the output current growth period information; Specifically, the actual growth period information is compared and analyzed with the output current growth period information to evaluate the accuracy of the growth period recognition model.

[0050] When the actual growth period information is inconsistent with the output current growth period information, the parameters of the growth period recognition model are optimized.

[0051] Specifically, if the actual growth period information is inconsistent with the output current growth period information, it means that the growth period recognition model may have errors or need to be updated, and the model parameters need to be optimized or the model needs to be retrained.

[0052] For example, it is assumed that the actual growth period of rice observed in the field is "late tillering period" and the actual height is 130 cm; this information will be input for subsequent analysis and comparison; the shooting height value extracted from the image data before is 120 cm, while the actual height value is 130 cm, so the first difference is 10 cm; assuming that the preset second threshold is 5 cm, since the first difference of 10 cm is greater than the second threshold, it is prompted that the video monitoring equipment needs to be adjusted, such as checking the angle of the camera, cleaning the lens, or adjusting the installation position of the equipment; assuming that the output current growth period information is "middle of the tillering period", while the actual growth period information is "late tillering period"; in this case, the two growth period information will be compared and analyzed to determine whether there is an error in the model; since the actual growth period information is inconsistent with the output current growth period information, the parameters of the growth period recognition model will be adjusted and optimized; this may involve adjusting the model's weights, learning rate and other parameters, or re-collecting more training data to retrain the model to improve the model's accuracy and generalization ability.

[0053] Alternatively, if Figure 4 As shown, the growth simulation method is used to generate and output the current growth period data and predicted height, including: Obtain historical growth days, historical height data, historical environmental factor data, and historical growth period information of rice to construct a second training set; Specifically, it is necessary to collect historical data of rice, including growth days, height data, environmental factor data (such as temperature, humidity, light, management measures, etc.) and corresponding growth period information; these data will be used to construct the second training set to provide a basis for the establishment of the growth simulation model; Taking the growth days, height data, and environmental factor data as inputs, and the growth period information of the next growth days as output, the second training set is used to establish and train the model to obtain a growth simulation model; Specifically, the growth days, height data, and environmental factor data are used as input features, and the growth period information of the next growth days is used as the output target. The second training set is used to establish and train a growth simulation model; this model can learn the relationship between rice growth and environmental factors, and predict the growth period information of a certain growth day in the future.

[0054] Obtain the current environmental factor data of rice, the current growing days, and the current predicted value of rice height to obtain the current data set; Specifically, we need to obtain the current environmental factor data of rice (such as temperature, humidity, light, fertilization, irrigation, etc.), the current growing days, and the current rice height prediction value obtained by the height prediction model. These data will constitute the current data set for the prediction of the growth simulation model; Inputting the current data set into the growth simulation model to obtain the growth period information of the next growth days, and inputting the information into a preset time series database; Specifically, the current data set is input into the growth simulation model to obtain the growth period information for the next growing days; this information will then be input into a preset time series database so that it can be quickly obtained when needed in the future.

[0055] Obtaining the current growth day, and obtaining the growth period information corresponding to the current growth day from the time series database; Specifically, the current data set can be used to calculate the growth period information of the next growing day and store it in the time series database; on the next growing day, the current growing day is obtained and the corresponding growth period information is retrieved from the time series database, thereby improving the efficiency of information acquisition and making full use of spare time to update the time series database.

[0056] Output the growth period information corresponding to the current growth days and the current predicted value of rice height.

[0057] For example, suppose that the growth data of a rice variety in the past year is collected, including the number of growing days, height measurements, ambient temperature, humidity, light intensity, and management measures such as fertilization and irrigation, and the growth period information of each time point (such as sowing period, tillering period, heading period, etc.) is recorded; a machine learning model suitable for processing time series data (such as long short-term memory network LSTM) is selected as the growth simulation model; the historical data is divided into a training set and a test set, and the training set is used to train the model, and the prediction error is minimized by continuously adjusting the parameters of the model; after the training is completed, the test set is used to evaluate the performance of the model to ensure that it can accurately predict the growth period information of the next number of growing days; assuming that the current date is a certain year, month, and day, through The current environmental factor data (such as temperature of 25°C, humidity of 70%, and light intensity of 5000 lux) are obtained through sensors, and the current rice height is predicted to be 120 cm through the height prediction model. These data will constitute the current data set; assuming that the growth simulation model predicts the growth period information of the next growing day as "mid-tillering period", this information is input into the time series database and associated with its corresponding growing day; assuming that the current growing day is the 30th day after sowing, the corresponding growth period information retrieved from the time series database is "early tillering period"; the growth period information corresponding to the current growing day (the 30th day after sowing) is output as "early tillering period", and the current rice height prediction value is 120 cm.

[0058] Alternatively, if Figure 4 As shown, the method of generating a rice growth report using the output data further includes: Enter the actual growth period information and actual height value; Calculating a second difference between the actual height value and the current predicted value of rice height; Determining whether the second difference is greater than a preset third threshold; If so, the parameters of the height prediction model are optimized.

[0059] Specifically, it is necessary to input the actual rice growth period information and the actual height value. This information is usually obtained through manual field observation or other reliable channels, and is used to compare and verify with the predicted data output by the system; calculate the difference between the actual height value and the current rice height prediction value, that is, the second difference, and this difference is used to evaluate the accuracy and stability of the height prediction model; determine whether the second difference is greater than a preset third threshold, which is set according to actual needs and model performance, and is used to determine whether the difference between the predicted value and the actual value is acceptable; if the second difference is greater than the preset third threshold, it means that the height prediction model may have errors or need to be updated. At this time, it is necessary to optimize the model parameters or retrain the model to improve the accuracy and generalization ability of the model.

[0060] For example, it is assumed that the actual growth period of rice is the "heading period" and the actual height is 150 cm according to artificial field observations. This information will be input for subsequent analysis and comparison. The current predicted value of rice height obtained by the height prediction model is 140 cm, while the actual height value is 150 cm, so the second difference is 10 cm. Assuming that the preset third threshold is 5 cm, since the second difference of 10 cm is greater than the third threshold, it is considered that the height prediction model has errors or needs to be adjusted. Since the second difference is greater than the third threshold, the parameters of the height prediction model will be adjusted and optimized, which may involve adjusting the model's weights, learning rate and other parameters, or re-collecting more training data to retrain the model. The optimized model will predict the height of rice more accurately. Through the above steps, the actual growth period information and the actual height value can be used to verify and adjust the predicted value of rice height to ensure the accuracy and reliability of the prediction results. At the same time, the height prediction model can be adjusted and optimized according to actual conditions to improve the performance and prediction ability of the model. Ultimately, this information will be used to generate a rice growth report to provide strong decision-making support for agricultural production.

[0061] Alternatively, if Figure 4 As shown, the determining whether the second difference is greater than a preset third threshold value further includes: If the second difference is less than or equal to the third threshold, determining whether the actual growth period information is consistent with the output growth period information corresponding to the current growth days; If they are inconsistent, the growth simulation model is optimized, and the time series database is updated using the optimized growth simulation model.

[0062] Specifically, determine whether the second difference (i.e., the difference between the actual height value and the current predicted rice height value) is less than or equal to the preset third threshold; if so, it means that the prediction result of the height prediction model is relatively close to the actual height value, and the accuracy of the model is within an acceptable range; next, it is necessary to determine whether the actual growth period information is consistent with the growth period information corresponding to the current growing days output by the growth simulation model. This step is to verify the accuracy of the growth simulation model in growth period prediction; if the actual growth period information is inconsistent with the growth period information corresponding to the current growing days output by the growth simulation model, it means that there is an error in the growth period prediction of the growth simulation model or it needs to be updated; at this time, it is necessary to optimize the parameters of the growth simulation model or retrain the model to improve the accuracy and generalization ability of the model; after optimizing the growth simulation model, it is necessary to use the optimized model to re-predict and update the growth period information in the time series database; in this way, the growth period information in the time series database will be more accurate and reliable, providing support for future predictions and decisions.

[0063] Exemplarily, assume that the calculated second difference is 3 cm, and the preset third threshold is 5 cm; since 3 cm is less than 5 cm, the prediction result of the height prediction model is considered to be relatively accurate; assume that the actually observed rice growth period is the "grain filling period", and the growth period predicted by the growth simulation model based on the current growing days is also the "grain filling period"; in this case, the actual growth period information is consistent with the model prediction result; assume that the actually observed rice growth period is the "grain filling period", but the growth period predicted by the growth simulation model is the "waxy period"; since the actual growth period information is inconsistent with the model prediction result, the parameters of the growth simulation model will be adjusted and optimized, such as adjusting the input features of the model, the optimization algorithm, etc. To improve the accuracy of the model in growing period prediction; the optimized growth simulation model is used to re-predict the rice growing period information in the future, and these prediction results are updated in the time series database; in this way, when it is necessary to query or predict the growing period information of a certain growing day in the future, more accurate and reliable information can be obtained directly from the time series database; through the above steps, not only the accuracy of the height prediction is paid attention to, but also the accuracy of the growing period information is ensured; when it is found that the growth simulation model has errors in the growing period prediction, the model can be optimized in time, and the optimized model can be used to update the time series database, thereby improving the prediction ability and decision support ability of the whole system.

[0064] Based on the same inventive concept, Figure 5 As shown, the present invention also provides a system for intelligently identifying rice growth based on video monitoring, the system comprising: A rice height acquisition module is used to obtain image data of rice using a video monitoring device, and to extract a shooting height value of the rice using the image data; A rice height prediction module is used to establish a height prediction model for predicting rice height, and use the height prediction model to predict rice height data at different time points to obtain a predicted height data set; A discrete degree evaluation module is used to evaluate the discrete degree of rice height by using the predicted height data set and the photographed height value to obtain an evaluation value; An evaluation value judgment module, used to judge whether the evaluation value is greater than a preset first threshold; An image recognition module, for extracting and outputting the current growth period data and the shooting height value by using an image recognition method if the evaluation value is greater than a first threshold value; A growth simulation module, for generating and outputting current growth period data and predicted height using a growth simulation method if the evaluation value is less than or equal to a first threshold; The growth report generation module is used to generate a rice growth report using the output data.

[0065] It should be noted that the electrical connection between the above-mentioned units does not necessarily mean direct connection of the lines, and the indirect connection mode can be applied to the embodiments of the present invention as long as the purpose of the present invention is achieved. The above is only an exemplary embodiment of the present invention and cannot be used to limit the scope of the present invention.

[0066] That is, any equivalent changes and modifications made according to the teachings of the present invention are still within the scope of the present invention. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present invention. This application is intended to cover any variation, use or adaptive change of the present invention, which follows the general principles of the present invention and includes common knowledge or customary technical means in the art that are not described in the present invention.

Claims

1. A method for intelligently identifying rice growth based on video monitoring, characterized in that: The method comprises: Acquiring image data of rice using video monitoring equipment, and extracting a shooting height value of the rice using the image data; Establishing a height prediction model for predicting rice height, and using the height prediction model to predict rice height data at different time points to obtain a predicted height data set; Using the predicted height data set and the photographed height value, evaluating the height dispersion of rice to obtain an evaluation value; Determining whether the evaluation value is greater than a preset first threshold; If so, the image recognition method is used to extract and output the current growth period data and the shooting height value; If not, the growth simulation method is used to generate and output the current growth period data and predicted height; Use the output data to generate a rice growth report.

2. The method for intelligently identifying rice growth based on video monitoring according to claim 1, characterized in that: The step of establishing a height prediction model for predicting the height of rice and using the height prediction model to predict the rice height data at different time points to obtain a predicted height data set includes: Collect historical rice growing days, historical environmental factor data, and historical growth heights corresponding to historical growing days to construct a historical data set; Taking the growth days and environmental factor data as input and the growth height as output, a neural network model is established and trained using the historical data set to obtain a height prediction model; The data of rice growth days and environmental factors are collected by day to obtain the input data set; Inputting the input data set into the height prediction model to obtain predicted values ​​of rice height corresponding to different growing days; The predicted values ​​of rice height corresponding to different growing days are integrated to obtain the predicted height data set.

3. The method for intelligently identifying rice growth based on video monitoring according to claim 1, characterized in that: The use of the predicted height data set and the photographed height value to evaluate the height dispersion of rice to obtain the evaluation value includes: Calculate the mean of the predicted values ​​of rice height using the predicted height data set; Calculate the standard deviation of the predicted value of rice height using the predicted height data set and the mean; Establish an evaluation function for evaluating the degree of dispersion of rice height, input the mean, the standard deviation and the shooting height value into the evaluation function, calculate and obtain an evaluation value, and for the evaluation value ,have , In the formula, is the shooting height value, is the mean of the predicted values ​​of rice height, is the standard deviation of the predicted value of rice height.

4. The method for intelligently identifying rice growth based on video monitoring according to claim 1, characterized in that: The method of extracting and outputting the current growth period data and the shooting height value by using the image recognition method includes: Acquire image data of rice at different growth stages from a preset rice growth stage image database to establish an image data set; Extracting feature data from image data of rice at different growth stages, and constructing a first training set using the feature data and the growth stages corresponding to the feature data; Taking the characteristic data as input and the growth period as output, a neural network model is established and trained using the first training set to obtain a growth period recognition model.

5. The method for intelligently identifying rice growth based on video monitoring according to claim 4, characterized in that: The method of extracting and outputting the current growth period data and the shooting height value by using an image recognition method also includes: Use video monitoring equipment to obtain current image data of rice; Extract features from the current image data of rice to obtain current feature data; Inputting the current characteristic data into the growth period recognition model to obtain the current growth period information; Output the current growth period information and the shooting height value.

6. The method for intelligently identifying rice growth based on video monitoring according to claim 5, characterized in that: The method of generating a rice growth report using the output data includes: Enter the actual growth period information and actual height value; Calculating a first difference between the actual height value and the photographed height value; When the first difference is greater than a preset second threshold, adjusting the video monitoring device; Comparing and analyzing the actual growth period information with the output current growth period information; When the actual growth period information is inconsistent with the output current growth period information, the parameters of the growth period recognition model are optimized.

7. The method for intelligently identifying rice growth based on video monitoring according to claim 1, characterized in that: The growth simulation method is used to generate and output the current growth period data and predicted height, including: Obtain historical growth days, historical height data, historical environmental factor data, and historical growth period information of rice to construct a second training set; Taking the growth days, height data, and environmental factor data as inputs, and the growth period information of the next growth days as output, the second training set is used to establish and train the model to obtain a growth simulation model; Obtain the current environmental factor data of rice, the current growing days, and the current predicted value of rice height to obtain the current data set; Inputting the current data set into the growth simulation model to obtain the growth period information of the next growth days, and inputting the information into a preset time series database; Obtaining the current growth day, and obtaining the growth period information corresponding to the current growth day from the time series database; Output the growth period information corresponding to the current growth days and the current predicted value of rice height.

8. The method for intelligently identifying rice growth based on video monitoring according to claim 7, characterized in that: The method of generating a rice growth report using the output data further comprises: Enter the actual growth period information and actual height value; Calculating a second difference between the actual height value and the current predicted value of rice height; Determining whether the second difference is greater than a preset third threshold; If so, the parameters of the height prediction model are optimized.

9. The method for intelligently identifying rice growth based on video monitoring according to claim 8, characterized in that: The determining whether the second difference is greater than a preset third threshold value further includes: If the second difference is less than or equal to the third threshold, determining whether the actual growth period information is consistent with the output growth period information corresponding to the current growth days; If they are inconsistent, the growth simulation model is optimized, and the time series database is updated using the optimized growth simulation model.

10. A system for intelligently identifying rice growth based on video monitoring, applying the method according to any one of claims 1 to 9, characterized in that: The system comprises: A rice height acquisition module is used to obtain image data of rice using a video monitoring device, and to extract a shooting height value of the rice using the image data; A rice height prediction module is used to establish a height prediction model for predicting rice height, and use the height prediction model to predict rice height data at different time points to obtain a predicted height data set; A discrete degree evaluation module is used to evaluate the discrete degree of rice height using the predicted height data set and the photographed height value to obtain an evaluation value; An evaluation value judgment module, used to judge whether the evaluation value is greater than a preset first threshold; An image recognition module, for extracting and outputting the current growth period data and the shooting height value by using an image recognition method if the evaluation value is greater than a first threshold value; A growth simulation module, for generating and outputting current growth period data and predicted height using a growth simulation method if the evaluation value is less than or equal to a first threshold; The growth report generation module is used to generate a rice growth report using the output data.

Images