Tomato growth analysis method and system based on image recognition
By combining tomato plant images and environmental data, using model cascades and specific models, the problem of low detection efficiency and accuracy of tomato disease in traditional methods is solved, and the rapid and accurate identification of tomato disease is achieved, reducing economic losses.
Patent Information
- Application Number
- CN202411029919.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-07-30
AI Technical Summary
Traditional image recognition methods have low classification efficiency and accuracy in tomato disease detection, making it difficult to achieve early warning, resulting in disease spread and economic losses.
By acquiring tomato plant image and environmental data, a tomato growth analysis and disease recognition model was established, and a model cascade and optimization module was adopted, combining ARIMA and Faster R-CNN models to identify disease locations and categories, reduce interference and improve identification speed.
It realizes rapid identification and accurate classification of tomato diseases, reduces economic losses, improves the recognition speed and accuracy of the model, and adapts to the growth pattern of four seasons.
Smart Images

Figure CN119007190A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image recognition, and in particular to a tomato growth analysis method and system based on image recognition. Background Art
[0002] Tomato has high nutritional value and good edibility. It is one of the important economic crops in my country. With the growth of food demand, the planting area and yield of tomatoes are increasing. However, tomatoes are easily infected by diseases. Disease infection can easily cause large-scale damage to plants, resulting in irreparable economic losses. Most tomato diseases start from the leaves and spread to the whole plant. Early warning of tomato leaf diseases can avoid the spread of diseases and reduce economic losses.
[0003] However, the traditional feature fusion method uses fixed weighted operations to enhance local features and suppress the expression of interfering features, but the differences in disease category images affect the generalization ability of the model, resulting in low classification efficiency and accuracy. Image classification technology is widely used in the agricultural field, especially in disease detection and classification, which is more efficient and accurate than traditional manual methods;
[0004] In view of the above technical defects, a solution is now proposed. Summary of the invention
[0005] The object of the present invention is to provide a tomato growth analysis method and system based on image recognition to solve the technical defects mentioned above. The present invention obtains a large number of tomato plant images and environmental data, and after pre-processing the data, models are respectively established for tomato growth analysis and disease identification. By cascading the acquired tomato plant disease identification model and tomato plant morphology identification model, the output of the tomato plant morphology recognition model is passed to the tomato plant disease identification model as input, and a quasi-model is generated and transmitted to the recognition optimization module to perform model cascade iteration, so that the model can quickly identify the location of tomato diseases and the category of pests and diseases to which they belong for labeling, and at the same time can better distinguish the interfering background in the tomato image in the image.
[0006] The object of the present invention can be achieved by the following technical solutions: a tomato growth analysis system based on image recognition, comprising a tomato image acquisition module, an image data processing module, a model building module, a data analysis and processing module, a model deployment module, an environmental data acquisition module and an identification optimization module;
[0007] The tomato image acquisition module is used to collect image information of tomato plants at different growth stages, the image information represents the morphology of tomato plants and leaf image data, and send the image information to the image data processing module;
[0008] The environmental data acquisition module is used to collect environmental data when the tomato image is collected. The environmental data includes the time when the tomato image is obtained, the external light intensity, weather and season information, and send the environmental data to the image data processing module;
[0009] After receiving the information from the tomato image acquisition module and the environmental data acquisition module, the image data processing module marks the location of the tomato disease and the category of the disease and pest in the image data, and sends it to the data analysis and processing module and the model building module;
[0010] After obtaining information from the image data processing module, the model building module obtains the tomato growth environment data and tomato plant morphology through the tomato growth analysis unit, and combines the external light intensity, weather and season information training set, adds the light intensity, weather and season information training set 2 on the basis of the ARIMA model, and obtains the tomato plant morphology recognition model; the disease recognition unit obtains the leaf image data set 1, and adds the detection of whether the tomato plant is affected by the disease and the corresponding pest and disease category on the basis of the Faster R-CNN model, and obtains the tomato plant disease recognition model;
[0011] After receiving the information from the image data processing module and the data from the model building module, the data analysis and processing module performs model cascading on the acquired tomato plant disease recognition model and tomato plant morphology recognition model. The model cascading arranges the tomato plant morphology recognition model and the tomato plant disease recognition model in sequence, wherein the output of the tomato plant morphology recognition model is passed to the tomato plant disease recognition model as input, and generates a quasi-model which is transmitted to the recognition optimization module;
[0012] After receiving the information from the data analysis and processing module, the recognition and optimization module adjusts the parameter configuration of the quasi-model and the parallel processing tasks, rewrites it into parallel or distributed mode, and transmits the quasi-model to the data analysis and processing module for iteration. After the quasi-model meets the required requirements, the quasi-model is output to the model deployment module.
[0013] After obtaining the data passed in by the recognition and optimization module, the model deployment module deploys the quasi-model, automatically processes the image, performs growth analysis, and generates corresponding reports.
[0014] Preferably, the image processing process of the image data processing module is as follows:
[0015] By acquiring the data of the tomato image acquisition module, cleaning the data, removing noise in the image, performing contrast adjustment and color correction on the image enhancement, dividing the image into background, tomato plant and tomato plant part, extracting the disease location area, marking the disease location and the category of the pest and disease to which it belongs, performing data enhancement on the leaf samples of the tomato mosaic disease category, and generating a tomato leaf pest and disease training set 1;
[0016] The receiving environment data acquisition module acquires the external environment data at the time of acquiring the tomato image, including: external light intensity, weather and season information, and combines it with the tomato plant morphology image to generate the second environment data training set.
[0017] Preferably, the establishment process of the disease identification unit in the model establishment module is as follows:
[0018] Faster R-CNN uses a regression network to adjust the position and size of the detection box. The coordinates of the initial anchor box are (x a ,y a ,w a ,h a ), the predicted offset is (t x ,t y ,t w ,t h ), w a is the width of the anchor box, h a is the height of the anchor box, x a is the x-axis value of the center coordinate of the anchor box, y a is the y-axis value of the center coordinate of the anchor box, t x The predicted value for the x-axis coordinate of the center of the anchor box, t y The predicted value of the y-axis coordinate of the center of the anchor box, t w is the predicted width of the anchor box, t h is the predicted height of the anchor box, x is the x coordinate of the center point of the real bounding box, Refers to t w An exponential function of , used to convert the predicted width into the scaling factor of the actual width, Refers to t h The exponential function is used to transform the predicted height into the scaling factor of the real height. The actual prediction box coordinates are obtained by the following formula. The specific formula is as follows:
[0019]
[0020] And move a fixed-size window on the image through the sliding window technology, and apply the model to each window for detection. The size of a tomato preprocessed image window is W×H, W is the window width, H is the window height, and it slides on the entire image. The specific formula is as follows:
[0021] W_{i,j}=I[i:i+W,j:j+H]
[0022] Where W_{i,j} represents the sub-window extracted from image I, where i and j are loop variables or index variables. On each sub-window, feature extraction and classifier are applied to detect the presence and status of tomatoes;
[0023] And through the convolution operation, a weighted summation operation is performed on the local area of the image, which is obtained by the following formula. The specific formula is as follows:
[0024]
[0025] Among them, I is the input image, K is the convolution kernel, (x, y) is the position of the output feature map, m and n are the radii of the convolution kernel K in the horizontal and vertical directions, respectively, and K(i, j) represents the weight value of the convolution kernel K at position (i, j), which extracts the shape of the leaves and the size of the leaf patches in the tomato image.
[0026] Preferably, the process of the tomato growth analysis unit is as follows:
[0027] The autocorrelation function ACF and the partial autocorrelation function PACF represent the correlation between the time series and its own lagged version. ACF represents
[0028]
[0029] Among them, γ k is the lag of the time series, k is the autocovariance, γ0 is the variance of the time series, and the non-stationary time series is transformed into a stationary time series through the difference operation in the ARIMA model, where the first-order difference is expressed as
[0030]
[0031] Among them, X t represents the observed value of the time series at time t, Indicates the observed value X at the current time t in the first-order difference operation t Subtract the observation value of the previous time t-1, and model the tomato growth analysis using seasonal differences and seasonal AR and MA orders. The seasonal part of the seasonal ARIMA model is expressed as
[0032]
[0033] Among them, (1-φ1B-…-φ P B P ), B is a lag operator, which represents the lag of the time series, (φ1,…,φ P ) is the seasonal autoregression coefficient, representing the seasonal lag effect of the time series, (1-Φ1B s -…-Φ Q B sQ ), B s represents the lag operator of the seasonal period s, (Φ1,...,Φ Q ) is the seasonal autoregressive coefficient, which is used to capture the long-term memory in the seasonal effect, Y t represents the seasonally adjusted time series value at time t, (1+θ1B+...+θ q B q ), (θ1,...,θ q ) is the seasonal moving average coefficient, which is used to deal with seasonal random fluctuations, (1+Θ1B s +...+Θ Q B sQ ), Θ1,...,Θ Q is the seasonal moving average coefficient, which is used to handle the long-term dependence of seasonality, Represents the white noise error term at time t, that is, the random fluctuation part not explained by the model.
[0034] Preferably, the analysis process of the data analysis processing module is as follows:
[0035] Cascade multiple convolutional layers and pooling layers to extract features from the image and gradually reduce the spatial size of the feature map. There is an image data of tomato plant morphology represented by X1, where each element represents a pixel value. The image is convolved using the convolution kernel K1 to obtain the feature map F1
[0036]
[0037] Among them, (i, j) is the pixel position in the feature map, (m, n) is the index of the convolution kernel, X1(i+m, j+n) is the pixel value in the image, K1(m, n) is the weight of the convolution kernel, which is used to reduce the spatial size of the feature map while retaining the main features. The feature map F1 is obtained after convolution.
[0038] Preferably, the optimization recognition process in the recognition optimization module is as follows:
[0039] By selecting tomato growth characteristics and pest and disease characteristics, the original disease identification data and tomato growth data are used to construct new original disease identification data and tomato growth data features. The prediction results of the model building module are combined, and the prediction results of different feature sets are fused through the model. The training data is increased by rotating, flipping, and scaling, and noise is added or new feature samples are generated to enrich the diversity of the data.
[0040] The beneficial effects of the present invention are as follows:
[0041] (1) The present invention uses tomato images and environmental data at the time of image acquisition, adds light intensity, weather and season information based on the ARI MA model, and models tomato growth analysis, so that the established model is more in line with the four-season growth law of tomato plants, so that the model has better periodic perception, and quickly identifies and characterizes tomatoes, thereby facilitating the identification of pest and disease areas on leaves and reducing the interference of image recognition caused by differences between tomato plants;
[0042] (2) The present invention cascades the acquired tomato plant disease recognition model and tomato plant morphology recognition model, and transmits the output of the tomato plant morphology recognition model to the tomato plant disease recognition model as input, thereby further reducing the overall recognition range of the model, reducing the model recognition time, and improving the model recognition speed. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] The present invention will be further described below in conjunction with the accompanying drawings;
[0044] Figure 1 is a reference diagram of the method of the present invention;
[0045] Figure 2 It is a system flow chart of the present invention. DETAILED DESCRIPTION
[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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] Embodiment 1:
[0048] See also Figure 1 to Figure 2 As shown, the present invention is a tomato growth analysis method based on image recognition, comprising the following steps:
[0049] Step 1: Acquire the tomato image data, and archive the environmental data information at the time of acquisition, and transfer the data to step 2, and correspond the tomato image data with the environmental data information at the time of acquisition;
[0050] Step 2: Obtain archived data, pre-process the data, summarize the environmental information corresponding to the image, and pass the data to steps 3 and 4 to enhance the data of the fewer tomato mosaic disease leaf samples;
[0051] Step 3: By obtaining the preprocessed data, model training is performed to establish a tomato growth analysis model and a disease identification model respectively, and the data is transferred to step 4, and the disease location area is extracted, and the disease location and the category of the pest are marked;
[0052] Step 4: The plant morphological labels and morphological images output by the tomato growth analysis model are input into the disease recognition model through cascading multiple layers to reduce the data processing time and scope, and the data is passed to step 5, cascading multiple convolutional layers and pooling layers to extract features in the image and gradually reduce the spatial size of the feature map;
[0053] Step 5: By selecting and adjusting the model parameters, combined with the prediction results of the model building module, the prediction results of different feature sets are integrated through the model, and the original disease identification data and tomato growth data are used to construct new original disease identification data and tomato growth data features, combined with the prediction results of the model building module, and the prediction results of different feature sets are integrated through the model;
[0054] A tomato growth analysis system based on image recognition, comprising a tomato image acquisition module, an image data processing module, a model building module, a data analysis and processing module, a model deployment module, an environmental data acquisition module and an identification optimization module;
[0055] The tomato image acquisition module is used to collect image information of tomato plants at different growth stages. The image information represents the morphology of tomato plants and leaf image data, and sends the image information to the image data processing module; the environmental data acquisition module is used to collect environmental data when tomato images are collected. The environmental data is the time when the tomato image is acquired, the external light intensity, weather and season information, and sends the environmental data to the image data processing module; after receiving the information from the tomato image acquisition module and the environmental data acquisition module, the image data processing module marks the location of tomato diseases and the category of pests and diseases in the image data, and sends it to the data analysis processing module and the model building module;
[0056] After obtaining information from the image data processing module, the model building module obtains the tomato growth environment data and tomato plant morphology through the tomato growth analysis unit, and combines the external light intensity, weather and season information training set, adds the light intensity, weather and season information training set 2 on the basis of the ARIMA model, and obtains the tomato plant morphology recognition model; the disease recognition unit obtains the leaf image data set 1, and adds the detection of whether the tomato plant is affected by the disease and the corresponding pest and disease category on the basis of the Faster R-CNN model, and obtains the tomato plant disease recognition model;
[0057] After receiving the information from the image data processing module and the data from the model building module, the data analysis and processing module performs model cascading on the acquired tomato plant disease recognition model and tomato plant morphology recognition model. The model cascading arranges the tomato plant morphology recognition model and the tomato plant disease recognition model in sequence, wherein the output of the tomato plant morphology recognition model is passed to the tomato plant disease recognition model as input, and generates a quasi-model which is transmitted to the recognition optimization module;
[0058] After receiving the information from the data analysis and processing module, the recognition and optimization module adjusts the parameter configuration of the quasi-model and parallelizes the processing tasks, rewrites it into parallel or distributed mode, and passes the quasi-model to the data analysis and processing module for iteration. After the quasi-model meets the required requirements, the quasi-model is output to the model deployment module; after obtaining the data passed by the recognition and optimization module, the model deployment module deploys the quasi-model, automatically processes the image and performs growth analysis, and generates the corresponding report; the image processing process of the image data processing module is as follows:
[0059] By acquiring the data of the tomato image acquisition module, the data is cleaned, the noise in the image is removed, the contrast adjustment and color correction are performed on the image enhancement, and the image is divided into the background, the tomato plant and the tomato plant part, and the disease location area is extracted, the disease location and the category of the pest and disease to which it belongs are marked, and the data of the leaf samples of the tomato mosaic disease category are enhanced, and the tomato leaf pest and disease training set 1 is generated;
[0060] The receiving environment data acquisition module acquires the external environment data at the time of tomato image acquisition, including: external light intensity, weather and season information, and combines it with the tomato plant morphological image to generate the second environmental data training set; the process of the tomato growth analysis unit is as follows: the autocorrelation function ACF and the partial autocorrelation function PACF represent the correlation between the time series and its own lagged version, respectively, and ACF represents
[0061]
[0062] Among them, γ kis the lag of the time series, k is the autocovariance, γ0 is the variance of the time series, and the non-stationary time series is transformed into a stationary time series through the difference operation in the ARIMA model, where the first-order difference is expressed as
[0063]
[0064] Among them, X t represents the observed value of the time series at time t, Indicates the observed value X at the current time t in the first-order difference operation t Subtract the observation value of the previous time t-1, and model the tomato growth analysis using seasonal differences and seasonal AR and MA orders. The seasonal part of the seasonal ARIMA model is expressed as
[0065]
[0066] Among them, (1-φ1B-…-φ P B P ), B is a lag operator, which represents the lag of the time series, (φ1,…,φ P ) is the seasonal autoregression coefficient, representing the seasonal lag effect of the time series, (1-Φ1B s -…-Φ Q B sQ ), B s represents the lag operator of the seasonal period s, (Φ1,…,Φ Q ) is the seasonal autoregressive coefficient, which is used to capture the long-term memory in the seasonal effect, Y t represents the seasonally adjusted time series value at time t, (1+θ1B+…+θ q B q ), (θ1,…,θ q ) is the seasonal moving average coefficient, which is used to deal with seasonal random fluctuations, (1+Θ1B s +…+Θ Q B sQ ), Θ1,…,Θ Q is the seasonal moving average coefficient, used to handle the long-term dependence of seasonality, t represents the white noise error term at time t, that is, the random fluctuation part not explained by the model;
[0067] Image information of tomato plants at different growth stages is collected. The image information represents the morphology of tomato plants and leaf image data, and the image information is sent to the image data processing module. Environmental data during tomato image collection is collected. The environmental data is the time when the tomato image is obtained, the external light intensity, weather and season information, and the environmental data is sent to the image data processing module. After receiving the information from the tomato image acquisition module and the environmental data acquisition module, the tomato disease location and the category of the disease and pest to which it belongs in the image data are marked and sent to the data analysis and processing module and the model building module. After the model building module obtains the information from the image data processing module, it obtains the tomato growth environment data and tomato plant morphology through the tomato growth analysis unit, and combines the external light intensity, weather and season information training set, and adds the light intensity, weather and season information training set 2 based on the ARIMA model to obtain the tomato plant morphology recognition model.
[0068] Embodiment 2:
[0069] The establishment process of the disease identification unit in the model establishment module is as follows:
[0070] Faster R-CNN uses a regression network to adjust the position and size of the detection box. The coordinates of the initial anchor box are (x a ,y a ,w a ,h a ), the predicted offset is (t x ,t y ,t w ,t h ), w a is the width of the anchor box, h a is the height of the anchor box, x a is the x-axis value of the center coordinate of the anchor box, y a is the y-axis value of the center coordinate of the anchor box, t x The predicted value for the x-axis coordinate of the center of the anchor box, t y The predicted value of the center coordinate y axis of the anchor box, t w is the predicted width of the anchor box, t h is the predicted height of the anchor box, x is the x coordinate of the center point of the real bounding box, Refers to t w An exponential function of , used to convert the predicted width into the scaling factor of the actual width, Refers to t h The exponential function is used to transform the predicted height into the scaling factor of the real height. The actual prediction box coordinates are obtained by the following formula. The specific formula is as follows:
[0071]
[0072] And move a fixed-size window on the image through the sliding window technology, and apply the model to each window for detection. The size of a tomato preprocessed image window is W×H, W is the window width, H is the window height, and it slides on the entire image. The specific formula is as follows:
[0073] W_{i,j}=I[i:i+W,j:j+H]
[0074] Where W_{i,j} represents a sub-window extracted from image I, where i and j are loop variables or index variables. On each sub-window, feature extraction and classifier are applied to detect the presence and status of tomatoes; and through convolution operation, weighted summation operation is performed on the local area of the image, which is obtained by the following formula. The specific formula is as follows:
[0075]
[0076] Among them, I is the input image, K is the convolution kernel, (x, y) is the position of the output feature map, m and n are the radii of the convolution kernel K in the horizontal and vertical directions respectively, K(i, j) represents the weight value of the convolution kernel K at the position (i, j), and the shape of the leaves and the size of the leaf patches in the tomato image are extracted; the analysis process of the data analysis and processing module is as follows: cascade multiple convolution layers and pooling layers to extract the features in the image and gradually reduce the spatial size of the feature map. There is an image data of the tomato plant morphology represented by X1, where each element represents a pixel value. The image is convolved using the convolution kernel K1 to obtain the feature map F1
[0077]
[0078] Among them, (i, j) is the pixel position in the feature map, (m, n) is the index of the convolution kernel, X1(i+m, j+n) is the pixel value in the image, K1(m, n) is the weight of the convolution kernel, which is used to reduce the spatial size of the feature map while retaining the main features. The feature map F1 obtained after convolution is subjected to the maximum pooling operation:
[0079] P(i,j)=max m,n∈pooling region F(i·s+m,j·s+n)
[0080] Among them, P is the feature map after pooling, and s is the step size of the pooling window;
[0081] The optimization recognition process in the recognition optimization module is as follows:
[0082] By selecting tomato growth characteristics and pest and disease characteristics, the original disease identification data and tomato growth data are used to construct new original disease identification data and tomato growth data features. The prediction results of the model building module are combined to fuse the prediction results of different feature sets through model fusion. The training data is increased by rotating, flipping, and scaling, and noise is added or new feature samples are generated to enrich the diversity of the data.
[0083] The disease recognition unit obtains leaf image dataset 1 and uses Faster On the basis of the R-CNN model, the detection of whether the tomato plant is attacked by disease and the corresponding pest and disease category are added to obtain the tomato plant disease recognition model. The obtained tomato plant disease recognition model and tomato plant morphology recognition model are cascaded. The model cascade arranges the tomato plant morphology recognition model and the tomato plant disease recognition model in sequence. As the depth of the network increases, by stacking multiple convolutional layers and pooling layers, each convolutional layer can increase the complexity and abstraction level of the features, while each pooling layer gradually reduces the size of the feature map while maintaining the expressive power of important features, allowing the network to learn more abstract and complex features at a higher level. At the same time, the spatial size of the feature map is reduced by the pooling layer, so that the entire network can process input images of different sizes and obtain good generalization ability. The output of the tomato plant morphology recognition model is passed to the tomato plant disease recognition model as input, and the quasi-model is passed to the data analysis and processing module for iteration. After the quasi-model meets the required requirements, the quasi-model is output to the model deployment module, the quasi-model is deployed, the image is automatically processed, the growth analysis is performed, and the corresponding report is generated.
[0084] The above formulas are obtained by collecting a large amount of data for software simulation and selecting a formula that is close to the actual value. The coefficients in the formula are set by technical personnel in this field according to actual conditions. The above are only preferred specific implementations of the present invention, but the protection scope of the present invention is not limited to this. Any technical personnel familiar with the technical field within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, make equivalent replacement or change, which should be covered within the protection scope of the present invention.
Claims
1. A tomato growth analysis system based on image recognition, characterized in that: It includes a tomato image acquisition module, an image data processing module, a model building module, a data analysis and processing module, a model deployment module, an environmental data acquisition module, and an identification and optimization module; The tomato image acquisition module is used to collect image information of tomato plants at different growth stages, the image information represents the morphology of tomato plants and leaf image data, and send the image information to the image data processing module; The environmental data acquisition module is used to collect environmental data when the tomato image is collected. The environmental data includes the time when the tomato image is obtained, the external light intensity, weather and season information, and send the environmental data to the image data processing module; After receiving the information from the tomato image acquisition module and the environmental data acquisition module, the image data processing module marks the location of the tomato disease and the category of the disease and pest in the image data, and sends it to the data analysis and processing module and the model building module; After obtaining information from the image data processing module, the model building module obtains the tomato growth environment data and tomato plant morphology through the tomato growth analysis unit, and combines the external light intensity, weather and season information training set, adds the light intensity, weather and season information training set 2 on the basis of the ARIMA model, and obtains the tomato plant morphology recognition model; the disease recognition unit obtains the leaf image data set 1, and adds the detection of whether the tomato plant is affected by the disease and the corresponding pest and disease category on the basis of the Faster R-CNN model, and obtains the tomato plant disease recognition model; After receiving the information from the image data processing module and the data from the model building module, the data analysis and processing module performs model cascading on the acquired tomato plant disease recognition model and tomato plant morphology recognition model. The model cascading arranges the tomato plant morphology recognition model and the tomato plant disease recognition model in sequence, wherein the output of the tomato plant morphology recognition model is passed to the tomato plant disease recognition model as input, and generates a quasi-model which is transmitted to the recognition optimization module; After receiving the information from the data analysis and processing module, the recognition and optimization module adjusts the parameter configuration of the quasi-model and the parallel processing tasks, rewrites it into parallel or distributed mode, and transmits the quasi-model to the data analysis and processing module for iteration. After the quasi-model meets the required requirements, the quasi-model is output to the model deployment module. After obtaining the data passed in by the recognition and optimization module, the model deployment module deploys the quasi-model, automatically processes the image, performs growth analysis, and generates corresponding reports.
2. A tomato growth analysis system based on image recognition according to claim 1, characterized in that: The image processing process of the image data processing module is as follows: By acquiring the data of the tomato image acquisition module, cleaning the data, removing noise in the image, performing contrast adjustment and color correction on the image enhancement, dividing the image into background, tomato plant and tomato plant part, extracting the disease location area, marking the disease location and the category of the pest and disease to which it belongs, performing data enhancement on the leaf samples of the tomato mosaic disease category, and generating a tomato leaf pest and disease training set 1; The receiving environment data acquisition module acquires the external environment data at the time of acquiring the tomato image, including: external light intensity, weather and season information, and combines it with the tomato plant morphology image to generate the second environment data training set.
3. The tomato growth analysis system based on image recognition according to claim 1, characterized in that: The establishment process of the disease identification unit in the model establishment module is as follows: Faster R-CNN uses a regression network to adjust the position and size of the detection box. The coordinates of the initial anchor box are (x a ,y a ,w a ,h a ), the predicted offset is (t x ,t y ,t w ,t h ), w a is the width of the anchor box, h a is the height of the anchor box, x a is the x-axis value of the center coordinate of the anchor box, y a is the y-axis value of the center coordinate of the anchor box, t x The predicted value for the x-axis coordinate of the center of the anchor box, t y The predicted value of the y-axis coordinate of the center of the anchor box, t w is the predicted width of the anchor box, t h is the predicted height of the anchor box, x is the x coordinate of the center point of the real bounding box, Refers to t w An exponential function of , used to convert the predicted width into the scaling factor of the actual width, Refers to t h The exponential function is used to transform the predicted height into the scaling factor of the real height. The actual prediction box coordinates are obtained by the following formula: And move a fixed-size window on the image through the sliding window technology, and apply the model to each window for detection. The size of a tomato preprocessed image window is W×H, W is the window width, H is the window height, and it slides on the entire image. The specific formula is as follows: W_{i,j}=I[i:i+W,j:j+H] Where W_{i,j} represents the sub-window extracted from image I, where i and j are loop variables or index variables. On each sub-window, feature extraction and classifier are applied to detect the presence and status of tomatoes; And through the convolution operation, a weighted summation operation is performed on the local area of the image, which is obtained by the following formula. The specific formula is as follows: Among them, I is the input image, K is the convolution kernel, (x, y) is the position of the output feature map, m and n are the radii of the convolution kernel K in the horizontal and vertical directions, respectively, and K(i, j) represents the weight value of the convolution kernel K at position (i, j), which extracts the shape of the leaves and the size of the leaf patches in the tomato image.
4. The tomato growth analysis system based on image recognition according to claim 1, characterized in that: The process of the tomato growth analysis unit is as follows: The autocorrelation function ACF and the partial autocorrelation function PACF respectively represent the correlation between the time series and its own lagged version. ACF is expressed as: Among them, γ k is the lag of the time series, k is the autocovariance, γ0 is the variance of the time series, and the non-stationary time series is transformed into a stationary time series through the difference operation in the ARIMA model, where the first-order difference is expressed as: ▽X t =X t -X t-1 Among them, X t represents the observed value of the time series at time t, ▽X t Indicates the observed value X at the current time t in the first-order difference operation t Subtract the observation at the previous time t-1 and model the tomato growth analysis using seasonal differences and seasonal AR and MA orders. The seasonal part of the seasonal ARIMA model is expressed as: (1-φ1B-...-φ P B P )(1-Φ1B s -...-Φ Q B sQ )Y t =(1+θ1B+...+θ q B q )(1+Θ1B s +...+Θ Q B sQ ) t Among them, (1-φ1B-...-φ P B P ), B is a lag operator, which represents the lag of the time series, (φ1,...,φ P ) is the seasonal autoregression coefficient, representing the seasonal lag effect of the time series, (1-Φ1B s -...-Φ Q B sQ ), B s represents the lag operator of the seasonal period s, (Φ1,...,Φ Q ) is the seasonal autoregressive coefficient, which is used to capture the long-term memory in the seasonal effect, Y t represents the seasonally adjusted time series value at time t, (1+θ1B+...+θ q B q ), (θ1,...,θ q ) is the seasonal moving average coefficient, which is used to deal with seasonal random fluctuations, (1+Θ1B s +...+Θ Q B sQ ), Θ1,...,Θ Q is the seasonal moving average coefficient, used to handle the long-term dependence of seasonality, t Represents the white noise error term at time t, that is, the random fluctuation part not explained by the model.
5. The tomato growth analysis system based on image recognition according to claim 1, characterized in that: The analysis process of the data analysis and processing module is as follows: Cascade multiple convolutional layers and pooling layers to extract features from the image and gradually reduce the spatial size of the feature map. There is an image data of tomato plant morphology represented by X1, where each element represents a pixel value. The image is convolved using the convolution kernel K1 to obtain the feature map F1: Among them, (i, j) is the pixel position in the feature map, (m, n) is the index of the convolution kernel, X1(i+m, j+n) is the pixel value in the image, K1(m, n) is the weight of the convolution kernel, which is used to reduce the spatial size of the feature map while retaining the main features. The feature map F1 is obtained after convolution.
6. The tomato growth analysis system based on image recognition according to claim 1, characterized in that: The optimization recognition process in the recognition optimization module is as follows: By selecting tomato growth characteristics and pest and disease characteristics, the original disease identification data and tomato growth data are used to construct new original disease identification data and tomato growth data features. Combined with the prediction results of the model building module, the prediction results of different feature sets are fused through the model, and the training data is increased by rotation, flipping, and scaling, and noise is added or new feature samples are generated.
7. A method for intelligent production scheduling based on big data, the system is applied to a tomato growth analysis system based on image recognition as described in any one of claims 1 to 6, characterized in that: The following steps are involved: Step 1: Acquire the tomato image data, archive and save the environmental data information at the time of acquisition, and transmit the data to step 2; Step 2: Obtain archived data, pre-process the data, summarize the environmental information corresponding to the image, and pass the data to steps 3 and 4; Step 3: By obtaining the preprocessed data, model training is performed to establish a tomato growth analysis model and a disease identification model, and the data is transferred to step 4; Step 4: Input the plant morphological labels and morphological images output by the tomato growth analysis model into the disease recognition model through cascading multiple layers, shortening the data processing time and scope, and passing the data to step 5; Step 5: By selecting and adjusting the model parameters, combined with the prediction results of the model building module, the prediction results of different feature sets are fused through the model.
Citation Information
Patent Citations
Deep learning-based tomato disease and insect pest detection and identification method
CN116630803A
Method for detecting plant diseases and insect pests of tomato leaves based on improved YOLOv5s
CN116994056A
An optimized approach for disease detection in tomato leaf using deep learning and internet of things.
IN202011028567A
