Wheat leaf disease monitoring method based on Internet of Things

Abstract

The invention relates to a wheat leaf disease monitoring method based on the Internet of Things. The method comprises the following steps: acquiring a wheat leaf image by using an Internet of Things camera; transmitting the wheat leaf images to an upper computer, and setting a corresponding disease type label for each wheat leaf image in the upper computer; performing illumination balance compensation on each wheat leaf image to generate a compensated wheat leaf image; inputting the compensated wheat leaf image as a sample into a neural network for training to obtain a wheat disease monitoring model; and inputting the target wheat leaf image into the wheat disease monitoring model to obtain the disease type of the corresponding wheat. According to the invention, the wheat leaf image is automatically collected by using the Internet of Things camera, and the wheat leaf image is input into the wheat disease monitoring model to complete disease identification, so that the dependence on manual inspection is reduced, the labor cost is reduced, and the wheat disease monitoring precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wheat disease monitoring, and in particular to a wheat leaf disease monitoring method based on the Internet of Things. Background Art

[0002] With the continuous growth of the global population, food security issues have become increasingly prominent, especially wheat, as one of the main food crops, whose yield and quality directly affect human survival and development. However, wheat is susceptible to various diseases during its growth, such as wheat powdery mildew, rust and leaf blight, which not only reduce wheat yield, but also affect its quality, causing serious economic losses to agricultural production.

[0003] Traditional wheat disease monitoring methods mainly rely on manual inspections and empirical judgments, which is not only time-consuming and labor-intensive, but also easily affected by subjective factors, making it difficult to detect and deal with diseases in the early stages. In addition, with climate change and changes in agricultural production methods, the occurrence patterns and transmission pathways of diseases have become more complex, and traditional monitoring methods are difficult to adapt to the needs of modern agriculture. Summary of the invention

[0004] To solve the above problems, an embodiment of the present invention aims to provide a wheat leaf disease monitoring method based on the Internet of Things.

[0005] A wheat leaf disease monitoring method based on the Internet of Things, comprising:

[0006] Step 1: Use the IoT camera to collect wheat leaf images;

[0007] Step 2: transmitting the wheat leaf image to a host computer, and setting a corresponding disease type label for each wheat leaf image in the host computer;

[0008] Step 3: Performing illumination balance compensation on each wheat leaf image to generate a compensated wheat leaf image;

[0009] Step 4: Input the compensated wheat leaf image as a sample into the neural network for training to obtain a wheat disease monitoring model;

[0010] Step 5: Input the target wheat leaf image into the wheat disease monitoring model to obtain the corresponding wheat disease type.

[0011] Preferably, the step 3: performing illumination balance compensation on each wheat leaf image to generate a compensated wheat leaf image comprises:

[0012] Step 3.1: Convert the wheat leaf image to the HSV color space, and extract the image of the brightness channel to form a brightness map;

[0013] Step 3.2: Extract the illumination component from the brightness map;

[0014] Step 3.3: Adaptively correct the pixels in the brightness map to form a corrected brightness map;

[0015] Step 3.4: Superimpose the corrected brightness map and the illumination component to form a compensated wheat leaf image.

[0016] Preferably, the step 3.2: extracting the illumination component from the brightness map comprises:

[0017] Step 3.2.1: Take a square guide window centered at any point in the brightness map;

[0018] Step 3.2.2: Use the Gaussian function to estimate the illumination component in the square guide window;

[0019] Step 3.2.3: Slide the square guide window to obtain the illumination component under each window, and weight it to obtain the illumination component of the brightness map; the illumination component calculation formula of the brightness map is:

[0020]

[0021] Among them, V_R (x,y) Represents the illumination component of the brightness map, w k represents the weight coefficient of the kth square guide window, n represents the number of square guide windows, I i(x,y) Represents the image within the square guide window, represents convolution, G (x,y) represents the Gaussian function, and σ represents the width of the Gaussian kernel.

[0022] Preferably, the step 3.3: adaptively correcting the pixels in the brightness map to form a corrected brightness map comprises:

[0023] Step 3.3.1: Use the sparrow search algorithm to search for the correction parameters of the gamma correction function to correct the pixels in the brightness map to obtain a preliminary corrected brightness map; wherein the gamma correction function is:

[0024]

[0025] Among them, φ represents the correction parameter, γ represents the enhancement coefficient, m represents the pixel mean in the brightness map, h represents the pixel point in the brightness map, and O(h) represents the brightness map after preliminary correction;

[0026] Step 3.3.2: Use the fitness function to evaluate the adaptability of the brightness map after preliminary correction. When the adaptability is less than a preset value, output the corrected brightness map.

[0027] Preferably, the step 4: inputting the compensated wheat leaf image as a sample into a neural network for training to obtain a wheat disease monitoring model comprises:

[0028] Step 4.1: Input the compensated wheat leaf image into a feature enhancement network to extract enhanced features; wherein the feature enhancement network includes a convolution layer and a feature enhancement extraction layer, and the feature enhancement extraction layer is used to sequentially perform depth convolution and point-by-point convolution on the extracted features; the feature enhancement network is used to fuse the features extracted by the convolution layer and the features extracted by the feature enhancement extraction layer to form enhanced features;

[0029] Step 4.2: Input the enhanced features as samples into the YOLOv3 network for training to obtain the wheat disease monitoring model.

[0030] Preferably, in step 4.1, the feature fusion formula of the feature enhancement network is:

[0031]

[0032] Where F represents the compensated wheat leaf image, F 100 Represents enhanced features, Conv 1×1 Represents a 1x1 convolutional layer, Conv 3×3 Represents a 3x3 convolutional layer, Conv 1×3 Represents a 1x3 convolutional layer, Conv 3×1 Represents a 3x1 convolutional layer, DWConv 3×3 represents the feature enhancement extraction layer, and Concat represents the concatenation operation.

[0033] Preferably, the step 4.2: inputting the enhanced features as samples into the YOLOv3 network for training to obtain a wheat disease monitoring model comprises:

[0034] Use multiple different types of loss functions to optimize the YOLOv3 network until the sum of multiple different types of loss functions is within the preset range, and output the wheat disease monitoring model; the optimization formula of the YOLOv3 network is:

[0035]

[0036] Among them, λ1 represents the first weight, λ2 represents the second weight, λ3 represents the third weight, L cla (O,C) represents the classification loss function, L cla (O,C) represents the confidence loss function, L cla (O, C) represents the positioning loss function, L(O, o, C, c, t, g) represents the sum of the loss functions, o iIndicates whether wheat leaves actually exist in the predicted bounding box i. represents the probability of whether there is a wheat leaf in the predicted bounding box i, O ij Indicates whether the predicted bounding box i actually contains wheat of the jth disease type. represents the probability that there is wheat of the jth disease type in the predicted bounding box i, (x i ,y i ) represents the center coordinate of the real bounding box, (ω i ,h i ) represents the width and height of the ground-truth bounding box, represents the center coordinates of the predicted bounding box, represents the width and height of the predicted bounding box, pos represents the number of bounding boxes, and cla represents the number of wheat disease types.

[0037] The present invention also discloses a wheat leaf disease monitoring system based on the Internet of Things, comprising:

[0038] An image acquisition module, used to acquire wheat leaf images using an IoT camera;

[0039] A disease type label setting module, used for transmitting the wheat leaf image to a host computer, and setting a corresponding disease type label for each wheat leaf image in the host computer;

[0040] A light balance module, used for performing light balance compensation on each wheat leaf image to generate a compensated wheat leaf image;

[0041] A training module, used for inputting the compensated wheat leaf image as a sample into a neural network for training to obtain a wheat disease monitoring model;

[0042] The disease type detection module is used to input the target wheat leaf image into the wheat disease monitoring model to obtain the corresponding wheat disease type.

[0043] The present invention also discloses an electronic device, comprising a bus, a transceiver, a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory and the processor are connected via the bus, and the computer program is characterized in that when the computer program is executed by the processor, the steps in the above-mentioned method for monitoring wheat leaf diseases based on the Internet of Things are implemented.

[0044] The present invention also discloses a computer-readable storage medium on which a computer program is stored, characterized in that when the computer program is executed by a processor, the steps in the above-mentioned method for monitoring wheat leaf diseases based on the Internet of Things are implemented.

[0045] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0046] The present invention relates to a wheat leaf disease monitoring method based on the Internet of Things. Compared with the prior art, the present invention automatically collects wheat leaf images by using an Internet of Things camera, and inputs the wheat leaf images into a wheat disease monitoring model to complete disease identification, which not only reduces the dependence on manual inspections and reduces labor costs, but also improves the accuracy of wheat disease monitoring.

[0047] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] 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 only 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.

[0049] Figure 1 A flow chart of a wheat leaf disease monitoring method based on the Internet of Things provided by the present invention;

[0050] Figure 2 This is a schematic diagram of the feature enhancement network provided by the present invention. DETAILED DESCRIPTION

[0051] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0052] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0053] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0054] See also Figure 1 , a wheat leaf disease monitoring method based on the Internet of Things, comprising:

[0055] Step 1: Use the IoT camera to collect wheat leaf images;

[0056] The wheat leaf images collected by the present invention are located in an agricultural demonstration base. Considering the changes in light intensity and angle during the day, which affect the accuracy of disease identification, the Internet of Things camera is selected to shoot and collect in the morning, noon, and evening. The camera height is 60cm-80cm away from the diseased leaves, and the image pixels are 1080P high-definition images. The types of diseased leaves include: wheat powdery mildew, wheat rust, wheat leaf blight, wheat gray mold, wheat leaf spot, etc.

[0057] Step 2: transmitting the wheat leaf image to a host computer, and setting a corresponding disease type label for each wheat leaf image in the host computer;

[0058] After the IoT camera collects and uploads enough types of diseased wheat leaves, the type of each diseased leaf needs to be manually labeled.

[0059] Step 3: Performing illumination balance compensation on each wheat leaf image to generate a compensated wheat leaf image;

[0060] In practical applications, uneven lighting may result in unclear color differences between normal leaves and diseased leaves. Therefore, after the present invention performs light balance compensation on each leaf, it can reduce misjudgment caused by lighting changes, enable the trained neural network model to better adapt to images under different lighting conditions, and enhance the robustness of the model.

[0061] Furthermore, step 3 includes:

[0062] Step 3.1: Convert the wheat leaf image to the HSV color space, and extract the image of the brightness channel to form a brightness map;

[0063] In practical applications, the human eye is more sensitive to brightness than to color, so the correction of the brightness component is the key to the illumination unevenness correction algorithm. For color images, if the correction is performed directly on the RGB three channels, it is difficult to ensure that each channel is enhanced or attenuated at the same ratio, which will cause color distortion in the image after correction, and the amount of calculation for processing the three channels at the same time is also relatively large. In view of the fact that the HSV color space is more in line with the visual characteristics of the human eye, and the hue (H), saturation (S) and brightness (V) in the HSV color space are independent of each other, the operation of the brightness V will not affect the color information of the image, so the present invention chooses to implement the correction of color images with uneven illumination in the HSV color space.

[0064] Step 3.2: Extract the illumination component from the brightness map;

[0065] Furthermore, step 3.2 includes:

[0066] Step 3.2.1: Take a square guide window centered at any point in the brightness map;

[0067] Step 3.2.2: Use the Gaussian function to estimate the illumination component in the square guide window;

[0068] Step 3.2.3: Slide the square guide window to obtain the illumination component under each window, and weight it to obtain the illumination component of the brightness map; the illumination component calculation formula of the brightness map is:

[0069]

[0070] Among them, V_R (x,y) Represents the illumination component of the brightness map, w k represents the weight coefficient of the kth square guide window, n represents the number of square guide windows, I i(x,y) Represents the image within the square guide window, represents convolution, G (x,y) represents the Gaussian function, and σ represents the width of the Gaussian kernel.

[0071] Gaussian filtering can help separate the illumination component in an image from the reflection component on the surface of an object. The present invention can obtain the illumination component of wheat leaves by performing Gaussian filtering weighted processing on pixels in a brightness map.

[0072] Step 3.3: Adaptively correct the pixels in the brightness map to form a corrected brightness map;

[0073] Furthermore, step 3.3 includes:

[0074] Step 3.3.1: Use the sparrow search algorithm to search for the correction parameters of the gamma correction function to correct the pixels in the brightness map to obtain a preliminary corrected brightness map; wherein the gamma correction function is:

[0075]

[0076] Among them, φ represents the correction parameter, γ represents the enhancement coefficient, m represents the pixel mean in the brightness map, h represents the pixel point in the brightness map, and O(h) represents the brightness map after preliminary correction;

[0077] Step 3.3.2: Use the fitness function to evaluate the adaptability of the brightness map after preliminary correction. When the adaptability is less than a preset value, output the corrected brightness map.

[0078] In order to verify the processing effect of the algorithm, the present invention uses the sum of objective indicators such as standard deviation (which can reflect the contrast characteristics of the image), average gradient (an important measure of image clarity) and entropy (a measure of the amount of information carried by the image) as a fitness function to measure the correction effect. When the adaptability is less than the preset value, it means that the correction effect of the present invention meets expectations.

[0079] Step 3.4: Superimpose the corrected brightness map and the illumination component to form a compensated wheat leaf image.

[0080] In step 3.4 of the present invention, the superposition formula of the present invention is: V_out=V_R+V_L, wherein V_out represents the value of the compensated V channel, and the compensated wheat leaf image can be formed by recombining the compensated V channel value with the hue (H) and saturation (S) of the original channel.

[0081] Step 4: Input the compensated wheat leaf image as a sample into the neural network for training to obtain a wheat disease monitoring model;

[0082] Furthermore, step 4 includes:

[0083] Step 4.1: Input the compensated wheat leaf image into a feature enhancement network to extract enhanced features; wherein the feature enhancement network includes a convolution layer and a feature enhancement extraction layer, and the feature enhancement extraction layer is used to sequentially perform depth convolution and point-by-point convolution on the extracted features; the feature enhancement network is used to fuse the features extracted by the convolution layer and the features extracted by the feature enhancement extraction layer to form enhanced features;

[0084] See also Figure 2In an embodiment of the present invention, the feature enhancement network of the present invention is roughly composed of a fusion of four branches. Among them, the first two branches are located in the shallow layer of the network, and extract local surface features, such as edges, colors, and textures, through 1×1 convolution operations. The third branch is located in the middle layer of the network. It further captures local edge texture information while reducing the amount of network parameters through 3x3 convolution and inverted bottleneck structure, and begins to gradually extract more complex shape features. Specifically, first, the input feature map is dimensionally expanded using 1x1 convolution, and the shallow features are mapped to a higher-dimensional space in order to capture and express the local features of the target in more dimensions. Subsequently, a 3×3 depthwise separable convolution is used to reduce the computational burden and further enhance the proposed features, while extracting deep shape features.

[0085] The fourth branch is located in the deep layer of the network, using 3×3 convolution and asymmetric convolution blocks to further strengthen the extracted local feature information. Specifically, this branch first uses a 3×3 ordinary convolution to strengthen the captured shallow local features, and then inputs these features into the asymmetric convolution block to extract shape features. The convolution block contains three parallel convolution layers: a 3×3 convolution layer, a 1x3 convolution layer, and a 3x1 convolution layer. Each convolution layer extracts features from different directions. Among them, the 3×3 convolution layer is used for all-round feature extraction, the 1x3 convolution layer focuses on feature extraction in the horizontal direction, and the 3x1 convolution layer focuses on feature extraction in the vertical direction. Finally, the features extracted by these three convolution layers are fully integrated to improve the accuracy of feature information.

[0086] Since the feature types and levels extracted by different branches are different, in order to obtain a more comprehensive and richer feature representation and reduce missed detection and false detection caused by the background confusion model, the present invention fuses these features. The final feature extraction process is as follows:

[0087]

[0088] Where F represents the compensated wheat leaf image, F 100 Represents enhanced features, Conv 1×1 Represents a 1x1 convolutional layer, Conv 3×3 Represents a 3x3 convolutional layer, Conv 1×3 Represents a 1x3 convolutional layer, Conv 3×1 Represents a 3x1 convolutional layer, DWConv 3×3 represents the feature enhancement extraction layer, and Concat represents the concatenation operation.

[0089] Step 4.2: Input the enhanced features as samples into the YOLOv3 network for training to obtain the wheat disease monitoring model.

[0090] In step 4.2, the present invention can use multiple different types of loss functions to optimize the YOLOv3 network until the sum of the multiple different types of loss functions is within a preset range, and output a wheat disease monitoring model; wherein the optimization formula of the YOLOv3 network is:

[0091]

[0092] Among them, λ1 represents the first weight, λ2 represents the second weight, λ3 represents the third weight, L cla (O,C) represents the classification loss function, L cla (O,C) represents the confidence loss function, L cla (O, C) represents the positioning loss function, L(O, o, C, c, t, g) represents the sum of the loss functions, o i Indicates whether wheat leaves actually exist in the predicted bounding box i. represents the probability of whether there is a wheat leaf in the predicted bounding box i, O ij Indicates whether the predicted bounding box i actually contains wheat of the jth disease type. represents the probability that there is wheat of the jth disease type in the predicted bounding box i, (x i ,y i ) represents the center coordinate of the real bounding box, (ω i ,h i ) represents the width and height of the ground-truth bounding box, represents the center coordinates of the predicted bounding box, represents the width and height of the predicted bounding box, pos represents the number of bounding boxes, and cla represents the number of wheat disease types.

[0093] The number of samples used for training and verification in the embodiment of the present invention is shown in Table 1:

[0094] Table 1

[0095] Disease type Number of images in the training set Number of validation set images Wheat powdery mildew 1351 212 Wheat rust 1254 206 Wheat leaf blight 1123 204 Wheat Botrytis cinerea 913 198 Wheat leaf spot 1034 203

[0096] The comparison of model accuracy is shown in Table 2:

[0097] Table 2

[0098]

[0099]

[0100] Step 5: Input the target wheat leaf image into the wheat disease monitoring model to obtain the corresponding wheat disease type.

[0101] The present invention can automatically collect wheat leaf images by utilizing the Internet of Things camera, and input the collected wheat leaf images into the wheat disease monitoring model through a wireless network to complete disease identification in real time, greatly reducing the dependence on manual inspections and reducing labor costs.

[0102] The present invention also discloses a wheat leaf disease monitoring system based on the Internet of Things, comprising:

[0103] An image acquisition module, used to acquire wheat leaf images using an IoT camera;

[0104] A disease type label setting module, used for transmitting the wheat leaf image to a host computer, and setting a corresponding disease type label for each wheat leaf image in the host computer;

[0105] A light balance module, used for performing light balance compensation on each wheat leaf image to generate a compensated wheat leaf image;

[0106] A training module, used for inputting the compensated wheat leaf image as a sample into a neural network for training to obtain a wheat disease monitoring model;

[0107] The disease type detection module is used to input the target wheat leaf image into the wheat disease monitoring model to obtain the corresponding wheat disease type.

[0108] Compared with the prior art, the beneficial effects of the wheat leaf disease monitoring system based on the Internet of Things provided by the present invention are the same as the beneficial effects of the wheat leaf disease monitoring method based on the Internet of Things described in the above technical scheme, which will not be repeated here.

[0109] The present invention also provides an electronic device, comprising a bus, a transceiver, a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory and the processor are connected via the bus, and wherein the computer program, when executed by the processor, implements the steps in the above-mentioned method for monitoring wheat leaf diseases based on the Internet of Things. Compared with the prior art, the beneficial effects of the electronic device provided by the present invention are the same as the beneficial effects of the method for monitoring wheat leaf diseases based on the Internet of Things described in the above-mentioned technical solution, and are not elaborated herein.

[0110] The present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps in the above-mentioned method for monitoring wheat leaf diseases based on the Internet of Things are implemented. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present invention are the same as the beneficial effects of the method for monitoring wheat leaf diseases based on the Internet of Things described in the above-mentioned technical scheme, which will not be repeated here.

[0111] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technical solution that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A wheat leaf disease monitoring method based on the Internet of Things, characterized in that: include: Step 1: Use the IoT camera to collect wheat leaf images; Step 2: transmitting the wheat leaf image to a host computer, and setting a corresponding disease type label for each wheat leaf image in the host computer; Step 3: Performing illumination balance compensation on each wheat leaf image to generate a compensated wheat leaf image; Step 4: Input the compensated wheat leaf image as a sample into the neural network for training to obtain a wheat disease monitoring model; Step 5: Input the target wheat leaf image into the wheat disease monitoring model to obtain the corresponding wheat disease type.

2. The method for monitoring wheat leaf diseases based on the Internet of Things according to claim 1, characterized in that: The step 3: performing illumination balance compensation on each wheat leaf image to generate a compensated wheat leaf image, comprises: Step 3.1: Convert the wheat leaf image to the HSV color space, and extract the image of the brightness channel to form a brightness map; Step 3.2: Extract the illumination component from the brightness map; Step 3.3: Adaptively correct the pixels in the brightness map to form a corrected brightness map; Step 3.4: Superimpose the corrected brightness map and the illumination component to form a compensated wheat leaf image.

3. The method for monitoring wheat leaf diseases based on the Internet of Things according to claim 2, characterized in that: The step 3.2: extracting the illumination component from the brightness map includes: Step 3.2.1: Take a square guide window centered at any point in the brightness map; Step 3.2.2: Use the Gaussian function to estimate the illumination component in the square guide window; Step 3.2.3: Slide the square guide window to obtain the illumination component under each window, and weight it to obtain the illumination component of the brightness map; the illumination component calculation formula of the brightness map is: Among them, V_R (x,y) Represents the illumination component of the brightness map, w k represents the weight coefficient of the kth square guide window, n represents the number of square guide windows, I i(x,y) Represents the image within the square guide window, represents convolution, G (x,y) represents the Gaussian function, and σ represents the width of the Gaussian kernel.

4. The method for monitoring wheat leaf diseases based on the Internet of Things according to claim 3, characterized in that: The step 3.3: adaptively correcting the pixels in the brightness map to form a corrected brightness map, includes: Step 3.3.1: Use the sparrow search algorithm to search for the correction parameters of the gamma correction function to correct the pixels in the brightness map to obtain a preliminary corrected brightness map; wherein the gamma correction function is: Among them, φ represents the correction parameter, γ represents the enhancement coefficient, m represents the pixel mean in the brightness map, h represents the pixel point in the brightness map, and O(h) represents the brightness map after preliminary correction; Step 3.3.2: Use the fitness function to evaluate the adaptability of the brightness map after preliminary correction. When the adaptability is less than a preset value, output the corrected brightness map.

5. The method for monitoring wheat leaf diseases based on the Internet of Things according to claim 4, characterized in that: The step 4: inputting the compensated wheat leaf image as a sample into a neural network for training to obtain a wheat disease monitoring model, including: Step 4.1: Input the compensated wheat leaf image into a feature enhancement network to extract enhanced features; wherein the feature enhancement network includes a convolution layer and a feature enhancement extraction layer, and the feature enhancement extraction layer is used to sequentially perform depth convolution and point-by-point convolution on the extracted features; the feature enhancement network is used to fuse the features extracted by the convolution layer and the features extracted by the feature enhancement extraction layer to form enhanced features; Step 4.2: Input the enhanced features as samples into the YOLOv3 network for training to obtain the wheat disease monitoring model.

6. The method for monitoring wheat leaf diseases based on the Internet of Things according to claim 5, characterized in that: In step 4.1, the feature fusion formula of the feature enhancement network is: Where F represents the compensated wheat leaf image, F 100 Represents enhanced features, Conv 1×1 Represents a 1x1 convolutional layer, Conv 3×3 Represents a 3x3 convolutional layer, Conv 1×3 Represents a 1x3 convolutional layer, Conv 3×1 Represents a 3x1 convolutional layer, DWConv 3×3 It represents the feature enhancement extraction layer, and Concat represents the concatenation operation.

7. A wheat leaf disease monitoring method based on the Internet of Things according to claim 6, characterized in that: The step 4.2: inputting the enhanced features as samples into the YOLOv3 network for training to obtain a wheat disease monitoring model, including: Use multiple different types of loss functions to optimize the YOLOv3 network until the sum of multiple different types of loss functions is within the preset range, and output the wheat disease monitoring model; the optimization formula of the YOLOv3 network is: Among them, λ1 represents the first weight, λ2 represents the second weight, λ3 represents the third weight, L cla (O,C) represents the classification loss function, L cla (O,C) represents the confidence loss function, L cla (O, C) represents the positioning loss function, L(O, o, C, c, t, g) represents the sum of the loss functions, o i Indicates whether wheat leaves actually exist in the predicted bounding box i. represents the probability of whether there is a wheat leaf in the predicted bounding box i, O ij Indicates whether the predicted bounding box i actually contains wheat of the jth disease type. represents the probability that there is wheat of the jth disease type in the predicted bounding box i, (x i ,y i ) represents the center coordinate of the true bounding box, (ω i ,h i ) represents the width and height of the ground-truth bounding box, represents the center coordinates of the predicted bounding box, represents the width and height of the predicted bounding box, pos represents the number of bounding boxes, and cla represents the number of wheat disease types.

8. A wheat leaf disease monitoring system based on the Internet of Things, characterized in that: include: An image acquisition module, used to acquire wheat leaf images using an IoT camera; A disease type label setting module, used for transmitting the wheat leaf image to a host computer, and setting a corresponding disease type label for each wheat leaf image in the host computer; A light balance module, used for performing light balance compensation on each wheat leaf image to generate a compensated wheat leaf image; A training module, used for inputting the compensated wheat leaf image as a sample into a neural network for training to obtain a wheat disease monitoring model; The disease type detection module is used to input the target wheat leaf image into the wheat disease monitoring model to obtain the corresponding wheat disease type.

9. An electronic device, comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, wherein: When the computer program is executed by the processor, the steps in the wheat leaf disease monitoring method based on the Internet of Things as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps in the wheat leaf disease monitoring method based on the Internet of Things as described in any one of claims 1 to 7 are implemented.

Images