A jujube tree disease and pest identification method and system combined with visual technology

Abstract

The application relates to the technical field of computer vision, in particular to a jujube tree disease and pest identification method and system combined with visual technology, which comprises the following steps: extracting a closed edge contour of a disease leaf gray image, and obtaining a contour skeleton of each closed edge contour; based on the graphic features of each contour skeleton, obtaining a to-be-identified region, analyzing the gradient amplitude distribution of the skeleton contour edge and the inside pixel points, the gradient amplitude distribution of the skeleton contour outside pixel points, and obtaining an edge transition value of each to-be-identified region; based on the overall distribution features of the gradient amplitude of all edge pixel points of the skeleton contour of each to-be-identified region, obtaining an edge energy value of each to-be-identified region; determining an edge distinguishing value of each to-be-identified region, and obtaining a disease identification result of each to-be-identified region. The application aims to improve the identification ability of jujube brown spot disease and jujube gray spot disease and improve the identification precision of disease detection.

Description

Technical Field

[0001] The present application relates to the field of computer vision technology, and in particular to a method and system for identifying pests and diseases of jujube trees in combination with vision technology. Background Art

[0002] Jujube tree diseases and pests have a significant negative impact on the growth, development, and economics of the jujube tree. They not only directly damage tree tissue and disrupt physiological metabolism, but also reduce fruit yield and quality. For example, jujube brown spot disease initially causes small brown spots on leaves, which expand into irregular dark brown lesions with yellow edges. In severe cases, the lesions become continuous, causing the leaves to wither and fall prematurely. Red spider mites, primarily adults and nymphs, infect the leaves, stems, flowers, and roots of plants, primarily targeting leaves. Infested leaves lose their green color and turn white, with dense, pale spots forming on the surface, curling and yellowing. In severe cases, leaves may turn yellow, burn, curl, fall, or even die. Jujube tree damage by pests and diseases hinders photosynthesis, leading to premature leaf and fruit drop, and consequently, significant reductions in agricultural production. Therefore, integrated pest and disease management is crucial, and timely prevention and control technologies for jujube tree diseases are essential to ensure the sustainable development of the jujube industry.

[0003] Real-time pest and disease detection in jujube areas using computer vision is a commonly used auxiliary technology for pest and disease control. The efficiency and real-time nature of computer vision detection minimizes the use of human resources. However, current methods for identifying jujube pests and diseases using visual technology can sometimes affect the detection capabilities of the visual algorithm due to the presence of different pests and diseases with similar characteristics, resulting in erroneous detection results and potentially affecting the prevention and control of the corresponding pests and diseases. Summary of the Invention

[0004] In view of the above, it is necessary to provide a method and system for identifying jujube tree diseases and pests combined with visual technology to solve the above problems.

[0005] The first aspect of the present application provides a method for identifying diseases and insect pests of jujube trees in combination with visual technology, the method comprising:

[0006] Collect grayscale images of diseased leaves of jujube trees;

[0007] Extract the closed edge contours of the grayscale image of the diseased leaf and obtain the contour skeleton of each closed edge contour; obtain the area to be identified based on the second-order center distance of each contour skeleton;

[0008] For the area to be identified, the distribution of the gradient amplitude of the skeleton contour edge and the inner pixel points, as well as the gradient amplitude distribution of the outer pixel points of the skeleton contour are analyzed to obtain the edge transition value of each area to be identified;

[0009] Based on the overall distribution characteristics of the gradient amplitude of all edge pixel points of the skeleton contour of each to-be-identified region, an edge energy value of each to-be-identified region is obtained, and in combination with an edge transition value, an edge distinguishing value of each to-be-identified region is determined, so as to obtain a disease identification result of each to-be-identified region.

[0010] Based on the disease identification result of each to-be-identified region, a deep learning model is trained to identify and detect diseases and pests of the jujube tree.

[0011] The to-be-identified region is obtained in the following manner:

[0012] The eccentricity of each contour skeleton is calculated using a second-order center distance, the eccentricity of all contour skeletons is threshold segmented to obtain an optimal threshold, and a region with an eccentricity higher than the optimal threshold is determined as a to-be-identified region.

[0013] The edge transition value of each to-be-identified region is obtained in the following manner:

[0014] For each to-be-identified region, a set composed of the gradient amplitudes of all pixel points on the skeleton contour edge and all pixel points adjacent to the skeleton contour by a first preset number of pixel units inside the skeleton contour is denoted as a first set.

[0015] A set composed of the gradient amplitudes of all pixel points adjacent to the skeleton contour by a second preset number of pixel units outside the skeleton contour edge is denoted as a second set.

[0016] The discrete degrees of all elements in the first set and the discrete degrees of all elements in the second set are compared to obtain an edge transition value of each to-be-identified region.

[0017] The edge transition value is specifically a result of positive fusion of a negative correlation mapping result of the discrete degrees of all elements in the second set and the discrete degrees of all elements in the first set.

[0018] The discrete degree is calculated using a variance.

[0019] The edge energy value of each to-be-identified region is obtained in the following formula: In the formula, Z represents the edge energy value of the to-be-identified region. represents a set of skeleton contour edge pixel points of the to-be-identified region. is a preset threshold value. represents the gradient amplitude of a pixel point a on the skeleton contour of the to-be-identified region; and e represents a natural constant.

[0020] The preset threshold value is determined by a preset quantile of a gradient amplitude histogram of the skeleton contour edge pixel points of the to-be-identified region.

[0021] The edge distinction value of each to-be-identified region is specifically a result of forward fusion of an edge energy value and an edge transition value.

[0022] The specific process of obtaining the disease identification result of each area to be identified is as follows:

[0023] Threshold segmentation is performed on the edge distinction values ​​of all areas to be identified to obtain a segmentation threshold, and areas with edge distinction values ​​greater than the segmentation threshold are identified as jujube gray spot areas, otherwise, they are identified as jujube brown spot areas.

[0024] In a second aspect, an embodiment of the present application also provides a date tree disease and pest identification system combined with visual technology, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above methods when executing the computer program.

[0025] This application has at least the following beneficial effects:

[0026] This application first extracts the closed edge contours of the grayscale image of the diseased leaf to obtain the contour skeleton of each closed edge contour; based on the second-order center distance of each contour skeleton, the area to be identified is obtained, which is helpful to extract the shape characteristics and structural information of the diseased area and enhance the subsequent accurate identification of the diseased area; the gradient amplitude of the edge pixel point can usually reflect the strength and clarity of the edge, and the gradient distribution of the inside and outside is helpful to distinguish the different stages or types of the disease, and the gradient amplitude distribution of the skeleton contour edge and the inner and outer pixel points is analyzed to obtain the edge transition value, which indicates how the edge of the diseased area transitions from the normal area, helps to identify the fuzziness or clarity of the regional boundary, and captures the significant differences between the diseased area and the healthy area; based on the overall distribution characteristics of the gradient amplitude of all edge pixel points of the skeleton contour of each area to be identified, the edge transition value is obtained. The edge energy value of each area to be identified is obtained, and the edge strength and structural complexity are quantified, which helps to determine the significance of the diseased area; the edge discrimination value of each area to be identified is determined, and the disease identification result of each area to be identified is obtained. The present application constructs the edge discrimination value by combining the edge transition value and the edge energy value. Taking into account that there are jujube brown spot disease and jujube gray spot disease with similar shapes and colors among the diseases that harm jujube tree leaves, conventional visual algorithms may make misjudgments during detection due to the similar characteristics of the two diseases. The present application distinguishes the two diseases from the edge feature distribution by combining the edge transition value and the edge energy value, uses the edge discrimination value to determine different disease areas and assists the subsequent algorithm to conduct targeted learning on the two different diseases, which can effectively improve the algorithm's ability to distinguish and detect similar diseases and improve the recognition accuracy of disease detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1A flowchart of a method for identifying pests and diseases of jujube trees using visual technology, provided in accordance with one embodiment of the present application;

[0028] Figure 2 A schematic diagram of calculating an edge distinction value provided in one embodiment of the present application. DETAILED DESCRIPTION

[0029] In the description of the embodiments of this application, words such as "exemplary," "or," and "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "or," and "for example" is intended to present the relevant concepts in a concrete manner.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art in the art of this application. The terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0031] It should also be noted that the terms "first" and "second" in this application and the accompanying drawings are used to distinguish similar objects, rather than to describe a specific order or sequence. The methods disclosed in the embodiments of this application or the methods shown in the flowcharts include one or more steps for implementing the methods. Without departing from the scope of protection of this application, the order of executing multiple steps can be interchanged with each other, and some steps can also be deleted.

[0032] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0033] The following describes in detail with reference to the accompanying drawings a specific scheme of a method and system for identifying pests and diseases of jujube trees combined with visual technology provided by this application.

[0034] See also Figure 1 , which shows a flowchart of a method for identifying pests and diseases of jujube trees combined with visual technology provided by an embodiment of the present application, the method comprising the following steps:

[0035] The first step: collect grayscale images of diseased leaves of jujube trees.

[0036] During the planting process of jujube trees, they are extremely susceptible to diseases and pests, which seriously affect their yield and quality. In the process of jujube tree growth and fruiting, jujube brown spot disease is the main leaf disease of jujube trees. Its pathogenic bacteria can cause a significant decrease in leaf photosynthetic efficiency, a decrease in fruit commercial rate and yield loss after infection. It has become an important biological factor restricting the sustainable development of the jujube industry. When combining visual technology to detect and prevent diseases of jujube trees, there is jujube gray spot disease, which is similar in color and shape to jujube brown spot disease among the diseases that harm jujube leaves. However, the prevention and control methods of the two diseases are completely different. Therefore, this application mainly focuses on distinguishing and judging the two similar diseases.

[0037] First, it is necessary to use a professional camera equipped with a high-definition camera to acquire images of leaves of jujube trees that are infected with diseases such as jujube brown spot disease and jujube gray spot disease. Since there may be a lot of environmental noise when acquiring images, it is necessary to perform denoising on the acquired images; this embodiment uses non-local mean denoising to denoise the acquired images. Non-local mean denoising can retain texture details and effectively remove Gaussian noise and speckle noise through weighted averaging of similar pixel blocks. At the same time, in order to reduce the amount of calculation of subsequent algorithms, the image is grayscale processed. In the embodiment of the present application, the grayscale processing method can be the maximum method; in other embodiments, the grayscale processing method can also use any one of the minimum method and the weighted average method to obtain the image of the diseased leaf.

[0038] The second step is to extract the closed edge contours of the grayscale image of the diseased leaf and obtain the contour skeleton of each closed edge contour; based on the second-order center distance of each contour skeleton, obtain the area to be identified.

[0039] When detecting and acquiring information about jujube brown spot, which can occur during the growth and fruiting of jujube trees, false positives are common due to the similarities between gray spot and brown spot. Gray spot and brown spot have different causes and require distinct control methods, requiring different medications for targeted disease prevention. Therefore, when testing for brown spot on jujube trees, false positives must be minimized.

[0040] When detecting and acquiring diseases, it is first necessary to use the Canny edge detection algorithm to extract the edge contour of the preprocessed diseased leaf image, obtain all closed contour areas in the diseased leaf image, and convert the images of all closed contour areas into binary images, that is, the grayscale value of all edge pixels is 255, which is a white foreground, and the grayscale value of the remaining pixels is 0, which is a black background; then the generated binary image is iteratively deleted using the Guo-Hall algorithm, and the output is a single-pixel wide, topology-preserving contour skeleton. Among them, the Guo-Hall algorithm is a well-known technology, and the specific calculation process will not be described in detail.

[0041] Among the diseases and pests of jujube tree leaves, jujube brown spot disease is different from other common diseases, such as dewdrop disease, jujube rust disease or red spider disease. The latter usually manifest as a nearly circular lesion area, while the lesions of jujube brown spot disease are irregular in shape. This irregular contour is obviously different from the circular lesion. The obtained contour skeleton coordinates are calculated using image moments, and the second-order center distance is output by calculation. The eccentricity of the contour skeleton is calculated using the second-order center distance. The eccentricity of all calculated contour skeletons is calculated using the Otsu threshold method to obtain the optimal threshold. The area with eccentricity lower than or equal to the optimal threshold is determined as a nearly circular area, and the area with eccentricity higher than the optimal threshold is determined as the area to be identified.

[0042] The third step: for the area to be identified, analyze the distribution of the gradient amplitude of the edge of the skeleton contour and the inner pixel points, and the gradient amplitude distribution of the outer pixel points of the skeleton contour to obtain the edge transition value of each area to be identified.

[0043] Among the diseases of jujube trees, brown spot disease of jujube bears similarities to gray spot disease in color and morphology. To avoid confusion during testing, further consideration is needed to distinguish the two diseases. Brown spot disease of jujube is caused by the spread of infection by pathogens (such as Alternaria alternata), the diffusion of toxins, and high humidity, which leads to a blurred boundary between diseased and healthy tissues. Initially, the lesions are halo-shaped, with edges becoming increasingly blurred as they expand rapidly. In contrast, gray spot disease has distinct edges, primarily due to the rapid killing of surrounding cells by pathogens (such as Phyllostomia jujuba) after infection, forming a distinct necrotic zone. Lesions expand more slowly, and the host plant produces defensive pigmentation at the junction of diseased and healthy tissues, creating a sharp boundary. Symptoms are more pronounced in low humidity. Based on the differences in the edges of the two diseased areas due to the different infected pathogens, the Sobel operator is used to obtain the gradient amplitudes of the edge skeleton contour pixel points, the pixel points on the inner side of the skeleton contour that are adjacent to the skeleton for a first preset number of pixel units, and the pixel points on the outer side of the skeleton contour that are adjacent to the skeleton for a second preset number of pixel units. In this embodiment, the first preset number is 1; the second preset number is 2.

[0044] Furthermore, the set consisting of the gradient amplitudes of all pixel points on the edge of the skeleton contour and all pixel points on the inner side of the skeleton contour that are adjacent to the skeleton contour by a first preset number of pixel units is recorded as the first set; the set consisting of the gradient amplitudes of all pixel points on the outer side of the skeleton contour that are adjacent to the skeleton contour by a second preset number of pixel units is recorded as the second set; the discrete degrees of all elements in the first set are compared with the discrete degrees of all elements in the second set to obtain the edge transition value of each area to be identified.

[0045] In this embodiment, the degree of discreteness of the set elements is calculated using variance; the degree of discreteness of all elements in the first set is recorded as A, and the degree of discreteness of all elements in the second set is recorded as B. Then the formula of the edge transition value is: ;in, It is a preset parameter. To prevent the denominator from being 0, the value is 0.01.

[0046] It should be understood that in the lesions of jujube brown spot disease, due to the infection of the pathogen, the edges of the diseased area show a halo effect. The diffusion of toxins causes the brightness changes on both sides of the lesion edge to be gradual rather than sudden. Therefore, the brightness gradient amplitude variance of the edge and internal pixels of the lesion is small, while the brightness of the pixels outside the lesion changes more significantly due to the transition from the diseased area to the healthy leaf, resulting in a relatively large variance in the peripheral area. Here, the gray-brown disease edge is clear, and the gradient amplitude of the edge contour is extremely high, while the inner necrosis is more uniform and the amplitude is lower. The gradient amplitude of the two circles of pixels is relatively discrete, so the numerator is large. The two circles of pixels outside the edge are the healthy part of the leaf, and the overall gradient amplitude is extremely low and close. The overall gradient amplitude is more concentrated, so the denominator is small, and the edge transition value is much greater than 1.

[0047] The fourth step: Based on the overall distribution characteristics of the gradient amplitude of all edge pixels of the skeleton contour of each area to be identified, the edge energy value of each area to be identified is obtained. Combined with the edge transition value, the edge distinction value of each area to be identified is determined to obtain the disease identification result of each area to be identified.

[0048] Considering that the core difference between jujube brown spot disease and jujube gray spot disease lies in the infection mechanism of the pathogen and the pathological response of the host plant, these biological differences directly lead to significant differences in the image edge performance of the two types of diseases. The toxins secreted by the brown spot pathogen require time to diffuse, resulting in the presence of "semi-necrotic" cells at the junction of diseased and healthy tissues, which is manifested as a gradual change in brightness. The overall gradient amplitude is moderate and dispersed. In contrast, the gray spot pathogen directly destroys the cell wall, causing "one-size-fits-all" necrosis. The brightness between diseased and healthy tissues suddenly changes, and the edge gradient amplitude is extremely high and concentrated. The gradient amplitude of all pixels in the entire area is obtained and formed into a histogram. The edge energy value Z is constructed by calculating the edge energy, and combined with the edge transition index as a disease differentiation index to distinguish the two diseases: Where, Represents a set of pixel points at the edge of the skeleton contour; is a preset threshold. In order to normalize the gradient strength, in this embodiment, high gradients are given high weights. is the 90% quantile of the edge gradient amplitude histogram; It represents the gradient amplitude of pixel a on the skeleton contour. The gradient amplitude is converted into an energy term through the exponential term, and the gradient amplitude is mapped to the interval (0,1]. The output of high gradient amplitude is close to 1, and the output of low gradient amplitude is close to 0. e represents a natural constant.

[0049] It should be understood that when the skeleton outline is jujube brown spot disease, the energy is dispersed due to the slow diffusion of the pathogen, and the calculated Z value is lower; when the skeleton outline is jujube gray spot disease, the energy is concentrated due to the rapid necrosis of the area, and the calculated Z value is higher.

[0050] Furthermore, an edge distinction value is generated by forward fusion of the edge energy value and the edge transition value. In this embodiment, a multiplication calculation method is used for forward fusion of multiple variables. It should be understood that when the area is jujube brown spot disease, since the edge energy value and the edge transition value are relatively low, the obtained edge distinction value will become further smaller; the edge distinction values ​​of all suspected jujube brown spot disease and jujube gray spot disease areas are calculated, and the Otsu threshold method is used to obtain the segmentation threshold, and the area with an edge distinction value higher than the segmentation threshold is determined to be a jujube gray spot disease area, and the area with an edge distinction value lower than or equal to the segmentation threshold is determined to be a jujube brown spot disease area.

[0051] The calculation diagram of the edge distinction value is as follows: Figure 2 shown.

[0052] The fifth step: Based on the disease identification results of each area to be identified, the deep learning model is trained to identify and detect diseases and pests in jujube trees.

[0053] After distinguishing the two diseased areas, the image of the captured area outlines was converted into a single-channel mask image with the background as 0, the jujube brown spot area as 1, and the jujube gray spot area as 2. Model training was performed using DeepLabv3+. The CBAM attention mechanism was added to the skip connection layer of the DeepLabv3+ network. By simultaneously optimizing both channel and spatial dimensions, the saliency of the diseased areas was enhanced, allowing for better learning of the distinguishing features of the diseased areas. A weighted hybrid loss of "primary loss + auxiliary loss" was used as the loss function to balance class imbalance and edge accuracy. The cross-entropy loss function was chosen as the primary loss function, and the Dice loss function was chosen as the auxiliary loss function. The AdamW optimizer was used, which uses adaptive momentum and regularization to suppress overfitting, making it suitable for a small-scale jujube leaf disease dataset. Training was performed for 300 epochs to prevent overfitting. A weighted model was obtained through training and then integrated into mobile monitoring equipment for model-assisted identification and detection of corresponding pests and diseases in jujube trees.

[0054] Based on the same inventive concept as the above-mentioned method, an embodiment of the present application also provides a jujube tree disease and pest identification system combined with visual technology, comprising a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-mentioned methods for identifying jujube tree diseases and pests combined with visual technology.

[0055] The flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to the embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the part of the module, program segment or code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in an order different from that disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified function or action, or may be implemented by a combination of dedicated hardware and computer instructions.

[0056] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the basic features of the present application. Therefore, from any point of view, the above embodiments of the present application should be regarded as exemplary and non-restrictive; modifications to the technical solutions described in the above embodiments, or equivalent replacement of some of the technical features therein, do not deviate from the essence of the corresponding technical solutions within the scope of the technical solutions of the embodiments of the present application, and should be included in the scope of protection of the present application.

Claims

1. A method for identifying diseases and insect pests of jujube trees using visual technology, characterized in that: The method comprises the following steps: Collect grayscale images of diseased leaves of jujube trees; The closed edge contours of the diseased leaf grayscale image are extracted to obtain the contour skeleton of each closed edge contour; based on the second-order center distance of each contour skeleton, the area to be identified is obtained; the area to be identified is the area with an eccentricity higher than the optimal threshold. The eccentricity of each contour skeleton is determined based on the second-order center distance, and the optimal threshold is obtained by threshold segmentation of the eccentricity of all contour skeletons; For the area to be identified, the distribution of the gradient amplitude of the edge and inner pixel points of the skeleton outline and the distribution of the gradient amplitude of the pixel points outside the skeleton outline are analyzed to obtain the edge transition value of each area to be identified; the edge transition value is obtained by comparing the discrete degree of all elements in a first set with the discrete degree of all elements in a second set, the first set being a set consisting of the gradient amplitudes of all pixel points on the edge of the skeleton outline and all pixel points inside the skeleton outline that are adjacent to the skeleton outline by a first preset number of pixel units, and the second set being a set consisting of the gradient amplitudes of all pixel points outside the skeleton outline that are adjacent to the skeleton outline by a second preset number of pixel units; Based on the overall distribution characteristics of the gradient amplitude of all edge pixels of the skeleton contour of each area to be identified, the edge energy value of each area to be identified is obtained. Combined with the edge transition value, the edge distinction value of each area to be identified is determined to obtain the disease identification result of each area to be identified; Based on the disease identification results of each area to be identified, the deep learning model is trained to identify and detect pests and diseases on jujube trees; The discrete degree of all elements in the first set is recorded as A, and the discrete degree of all elements in the second set is recorded as B. The formula of the edge transition value is: ;in, It is a preset parameter. To prevent the denominator from being 0, the value is 0.01; Get the edge energy value of each area to be identified. The specific formula is: ; Where Z represents the edge energy value of the area to be identified; A set of pixel points representing the edge of the skeleton contour of the area to be identified; is the preset threshold; It represents the gradient amplitude of pixel point a on the skeleton contour of the area to be identified; e represents a natural constant.

2. The method for identifying pests and diseases of jujube trees in combination with visual technology according to claim 1, characterized in that: The degree of dispersion is calculated using variance.

3. The method for identifying pests and diseases of jujube trees in combination with visual technology according to claim 1, characterized in that: The preset threshold is determined by a preset quantile of a gradient amplitude histogram of edge pixel points of the skeleton contour of the area to be identified.

4. The method for identifying pests and diseases of jujube trees in combination with visual technology according to claim 1, characterized in that: The edge distinction value of each to-be-identified region is specifically a result of forward fusion of the edge energy value and the edge transition value.

5. The method for identifying pests and diseases of jujube trees in combination with visual technology according to claim 1, characterized in that: The specific process of obtaining the disease identification result of each area to be identified is: Threshold segmentation is performed on the edge distinction values ​​of all areas to be identified to obtain a segmentation threshold, and areas with edge distinction values ​​greater than the segmentation threshold are identified as jujube gray spot areas, otherwise, they are identified as jujube brown spot areas.

6. A jujube tree pest and disease identification system combined with visual technology, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

Images