Camellia oleifera young forest growth monitoring system and method based on deep learning

Abstract

The application discloses a tea-oil camellia young forest growth monitoring system and method based on deep learning, and relates to the technical field of agricultural monitoring, comprising: acquiring image data and ground environment sensing data, generating a graph data structure containing tea-oil camellia young plant nodes, intercropping crop nodes and corresponding edge relationships, and binding the current height value of the tea-oil camellia young plant nodes, then extracting the corresponding ecological modeling function according to the edge relationship, combining the environment sensing data to calculate the dynamic edge weight value, and updating the weight of each edge in the graph data structure with the dynamic edge weight value, then inputting the updated graph structure data into a pre-trained graph neural network model, outputting the theoretical height value of each tea-oil camellia young plant after removing the interference of intercropping at the current time point, and determining whether the corresponding tea-oil camellia young plant is abnormal according to the theoretical height value; the beneficial effects are that the resource competition relationship between intercropping crops and tea-oil camellia young plants can be quantified, the growth abnormality can be accurately identified, and the misjudgment rate can be reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural monitoring, and in particular to a camellia oleifera young forest growth monitoring system based on deep learning and a method thereof. BACKGROUND

[0002] Camellia oleifera is an important woody oil crop with high economic and ecological benefits. Due to the long growth cycle and slow fast-growing period of camellia oleifera, especially in the young forest stage, the quality of its cultivation and management directly determines the future oil production capacity and forest quality. Therefore, accurate monitoring of camellia oleifera young forest has important practical significance.

[0003] Existing forest growth monitoring technologies mainly rely on unmanned aerial vehicle visible light images, multispectral imaging or ground image acquisition, combined with deep learning algorithms for image segmentation, target recognition and growth trend evaluation. Such technologies can efficiently monitor forest canopy expansion, plant density changes and early symptoms of diseases and pests in a large area, and have the advantages of convenient operation, high information density and strong applicability. However, for camellia oleifera young forest, in order to fully utilize the land, an interplanting cultivation mode is usually implemented, that is, other interplanting crops are planted between camellia oleifera young plants. Under this planting mode, the growth speed of different camellia oleifera young plants is greatly different due to the influence of adjacent interplanting crop types. If only image is used for evaluation, it is difficult to distinguish abnormal individuals from camellia oleifera young plants with different growth speeds, which may lead to the situation that "normal" but actually abnormal young plants are neglected for maintenance, or "abnormal" but actually normal young plants are over-maintained.

[0004] Therefore, a camellia oleifera young forest growth monitoring system and method based on deep learning are proposed. SUMMARY

[0005] In view of the above existing technical conditions, the present application is proposed. The embodiments of the present application provide a camellia oleifera young forest growth monitoring system and method based on deep learning, which can quantify the resource competition relationship between interplanting crops and camellia oleifera young plants, accurately identify growth abnormalities and reduce the misjudgment rate.

[0006] According to an aspect of the present application, a deep learning-based camellia sapling growth monitoring method is provided, comprising: acquiring image data and ground environment sensing data covering a target camellia sapling area; generating a graph data structure corresponding to the target camellia sapling area and current height values of each camellia sapling according to the image data and the environment sensing data, the graph data structure comprising camellia sapling nodes, intercropping crop nodes and corresponding edge relationships, the camellia sapling nodes being bound to the current height values; extracting corresponding ecological modeling functions from a preset relationship-function mapping table according to the edge relationships; calculating dynamic edge weights reflecting resource competition relationships between each camellia sapling and adjacent intercropping crops according to the ecological modeling functions and the environment sensing data; updating the weights of each edge in the graph data structure according to the dynamic edge weights; inputting the updated graph structure data into a pre-trained graph neural network model to output theoretical height values of each camellia sapling in the target camellia sapling area after removing the influence of intercropping interference at a current time point; and determining that the corresponding camellia sapling is growing abnormally if the theoretical height values are less than a preset acceptable height value.

[0007] According to another aspect of the present application, a deep learning-based camellia sapling growth monitoring system is provided, comprising: a data acquisition module for acquiring image data and ground environment sensing data covering a target camellia sapling area; a graph generation module for generating a graph data structure corresponding to the target camellia sapling area and current height values of each camellia sapling according to the image data and the environment sensing data, the graph data structure comprising camellia sapling nodes, intercropping crop nodes and corresponding edge relationships, the camellia sapling nodes being bound to the current height values; a function extraction module for extracting corresponding ecological modeling functions from a preset relationship-function mapping table according to the edge relationships; an edge weight calculation module for calculating dynamic edge weights reflecting resource competition relationships between each camellia sapling and adjacent intercropping crops according to the ecological modeling functions and the environment sensing data; a graph update module for updating the weights of each edge in the graph data structure according to the dynamic edge weights; a growth prediction module for inputting the updated graph structure data into a pre-trained graph neural network model to output theoretical height values of each camellia sapling in the target camellia sapling area after removing the influence of intercropping interference; and a decision module for determining that the corresponding camellia sapling is growing abnormally if the theoretical height values are less than a preset acceptable height value.

[0008] According to another aspect of the present application, an electronic device is provided, comprising a memory for storing computer executable instructions and a processor for executing the computer executable instructions, the computer executable instructions being executed by the processor to implement the steps of the method described above.

[0009] According to another aspect of the present application, a computer storage medium is provided, having stored thereon computer-executable instructions that, when executed by a processor, implement the steps of the method as described above.

[0010] Compared with the prior art, the tea-oil camellia young forest growth monitoring system and method based on deep learning according to the embodiments of the present application can effectively quantify the hidden ecological influence of intercropping crops on tea-oil camellia young trees by constructing a graph data structure containing ecological competition relationships and dynamically updating edge weights, and combining a graph neural network model to predict the theoretical growth height without the interference of intercropping, and has the advantages of accurately identifying growth abnormalities and reducing the misjudgment rate by quantifying the resource competition relationship between intercropping crops and tea-oil camellia young trees. BRIEF DESCRIPTION OF DRAWINGS

[0011] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description of the preferred embodiments of the present application taken in conjunction with the accompanying drawings. The drawings provided herein are for illustrative purposes only and, therefore, are not to be construed as being to scale. The following figures are provided to further illustrate the embodiments of the present application and are not to be construed as limiting the scope of the present application.

[0012] Figure 1 Flow chart of the tea-oil camellia young forest growth monitoring method based on deep learning of the present application.

[0013] Figure 2 Block diagram of the tea-oil camellia young forest growth monitoring system based on deep learning of the present application.

[0014] Figure 3 Block diagram of an electronic device of the present application. DETAILED DESCRIPTION

[0015] Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are merely a part of the embodiments of the present application, but not the whole embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.

[0016] Exemplary method

[0017] Figure 1The method for monitoring growth of camellia sinensis seedling forest based on deep learning according to the embodiment of the application is illustrated, which comprises: acquiring image data and ground environment sensing data covering a target camellia sinensis seedling forest area; generating a graph data structure corresponding to the target camellia sinensis seedling forest area and current height values of each camellia sinensis seedling according to the image data and the environment sensing data, the graph data structure comprising camellia sinensis seedling nodes, intercropping crop nodes and corresponding edge relationships, and the camellia sinensis seedling nodes being bound to the current height values; extracting corresponding ecological modeling functions from a preset relationship-function mapping table according to the edge relationships; calculating dynamic edge weights reflecting resource competition relationships between each camellia sinensis seedling and adjacent intercropping crops according to the ecological modeling functions and the environment sensing data; updating the weights of each edge in the graph data structure according to the dynamic edge weights; inputting the updated graph structure data into a pre-trained graph neural network model to output theoretical height values of each camellia sinensis seedling in the target camellia sinensis seedling forest area at a current time point after removing the influence of intercropping interference; and judging whether the theoretical height values are less than a preset acceptable height value, and if so, determining that the corresponding camellia sinensis seedling is growing abnormally.

[0018] Among them, the image data covering the target camellia sinensis seedling forest area refers to visible light images and multispectral images collected by unmanned aerial vehicles or ground camera equipment, and specifically, a target detection algorithm can be used to identify the spatial distribution of camellia sinensis seedlings and intercropping crops. The ground environment sensing data refers to real-time parameters obtained by soil moisture sensors, light intensity sensors, temperature and humidity sensors, laser radars, etc., and specifically, Internet of Things devices can be used to periodically collect and transmit the data to cloud servers.

[0019] Among them, the graph data structure refers to the topological relationship between camellia sinensis seedlings and intercropping crops expressed in the form of graph theory.

[0020] Among them, the graph neural network model refers to a deep learning model for feature learning based on graph structure data, and specifically, a graph attention network or a graph convolution network can be used to predict the theoretical growth state after removing the interference of intercropping by aggregating node features and edge weight information.

[0021] Among them, the theoretical height value refers to the expected growth height of camellia sinensis seedlings after excluding the competition effect of intercropping crops, and specifically, it can be obtained by regression analysis of historical growth data and competition factors through the graph neural network model. By eliminating the influence of intercropping crops on all camellia sinensis seedlings under intercropping mode, the individual with abnormal growth can be accurately determined.

[0022] The core innovation of the application lies in constructing a dynamic graph structure integrating multiple sources of data, quantifying the competition effect of intercropping crops on camellia sinensis seedlings through ecological modeling functions, and predicting the theoretical growth value after removing the interference based on the graph neural network, so as to accurately identify the growth abnormalities caused by resource competition. This method breaks through the limitation of traditional image monitoring technology that cannot distinguish between natural growth differences and intercropping interference, and realizes accurate diagnosis of the growth state of camellia sinensis seedling forest.

[0023] Through the above scheme, the application can effectively solve the problem that the traditional camellia sinensis young forest monitoring method cannot accurately identify abnormal growth individuals under intercropping mode. By constructing a dynamic atlas structure and combining a graph neural network model, the method can accurately quantify the complex ecological competition relationship between camellia sinensis seedlings and intercropping crops, thereby accurately evaluating the actual growth conditions of each camellia sinensis seedling. This method not only can distinguish between natural growth differences and growth abnormalities caused by intercropping interference, but also can adapt to changes in different intercropping crop combinations and planting densities, greatly improving the accuracy and applicability of camellia sinensis young forest growth monitoring. By identifying abnormal plants that truly need intervention in a timely manner, the method helps to optimize the allocation of maintenance resources, improve the survival rate of young forests and the quality of forestation, thereby laying the foundation for improving camellia sinensis yield and ecological benefits.

[0024] In some of the above schemes of the application, the ecological modeling function includes at least one of the following functions: a water competition function for representing the relationship between soil water absorption of camellia sinensis seedlings and intercropping crops; a allelopathy inhibition diffusion function for depicting the influence of allelochemicals released by intercropping crops on camellia sinensis seedlings; and a light shading index function for quantifying the light attenuation intensity caused by intercropping crops shading camellia sinensis seedlings.

[0025] Specifically, the water competition function is:

[0026] ;

[0027] wherein, represents the water competition coefficient between camellia sinensis seedling i and intercropping crop j, represents the root water absorption capacity of crop j per unit time, represents the root coverage radius of intercropping crop j, represents the horizontal distance between camellia sinensis seedling i and crop j, is a preset minimum value set to avoid division by zero error.

[0028] wherein, the root water absorption capacity is calibrated by soil moisture sensor and root distribution model, the root coverage radius is obtained by matching and acquiring from a preset crop parameter library according to the type of intercropping crop. The horizontal distance is obtained by calculating the Euclidean distance from the plant position coordinates in the image data, and the minimum value is set to 1e-6 to avoid numerical errors when the denominator is zero. The water competition coefficient and and are positively correlated, and is inversely proportional to the square, reflecting the combined effect of the water absorption capacity, root range and spatial distance of intercropping crops on water competition.

[0029] Specifically, after generating the graph data structure, the competition intensity between each pair of camellia sapling and adjacent intercropping crop is calculated by the water competition function. When the root water absorption capacity of the intercropping crop is stronger, the coverage radius is larger, and the distance from the camellia sapling is closer, the water competition coefficient significantly increases. This coefficient is input as a dynamic edge weight value into the graph neural network, enabling the model to quantify the water deprivation effect of intercropping crops on camellia saplings. For example, when corn is used as an intercropping crop, its developed fibrous root system leads to a higher value, which will result in a larger water competition coefficient at the same distance, thereby more accurately correcting the theoretical height prediction value during the graph neural network inference process and avoiding real growth abnormalities masked by water competition.

[0030] Specifically, the allelopathic inhibition diffusion function is:

[0031] ;

[0032] wherein, represents the allelopathic inhibition intensity of intercropping crop j on camellia sapling i, represents the concentration coefficient of allelochemicals released by intercropping crop j per unit time, represents the horizontal distance between camellia sapling i and crop j, represents the diffusion attenuation coefficient of allelochemicals released by intercropping crop j in the soil.

[0033] wherein, is obtained through laboratory determination or historical data fitting, reflecting the potential ability of different intercropping crops to release allelochemicals. Based on the calculation of plant position coordinates in image data, the accuracy of spatial distance is ensured. λ is dynamically adjusted according to soil type, humidity, and microbial activity parameters, for example, set to 0.15 in clay and 0.25 in sandy soil, to match the actual diffusion rate of allelochemicals in different media. The exponential function is used to simulate the attenuation law of allelochemical concentration with distance, and when increases, the inhibition effect decreases nonlinearly.

[0034] Specifically, when calculating the allelopathic inhibition intensity of camellia sapling i and intercropping crop j, first obtain the allelochemical concentration coefficient of intercropping crop j , combine the horizontal distance between them and the preset diffusion attenuation coefficient λ, and substitute into the exponential function to obtain . For example, when the intercropping crop is soybean, the value is 2.3, if is 1.5 meters, and λ is 0.2, then =2.3·exp(-0.2·1.5)=2.3·0.7408≈1.704. This value is input into the graph neural network as a dynamic edge weight to correct the theoretical height prediction of the young camellia oleifera plant.

[0035] Specifically, the light shading index function is:

[0036] ;

[0037] wherein, represents the light shading influence intensity of intercropping crop j on young camellia oleifera plant i, represents the canopy density index of intercropping crop j, represents the unit leaf area index of intercropping crop j, represents the horizontal distance between young camellia oleifera plant i and crop j, represents the current height difference between intercropping crop j and young camellia oleifera plant i, represents the sensitivity coefficient of vertical height difference to shading degree, represents the light shading weight factor, a minimum constant set to avoid division by zero error.

[0038] wherein, the canopy density index is obtained by analyzing the branch and leaf coverage density of intercropping crop j through multispectral image, the unit leaf area index is calculated based on leaf morphological characteristics and the number of leaves per unit area, the horizontal distance is calculated by the Euclidean distance through image positioning coordinates, and the height difference is measured by laser radar point cloud data or stereo vision between intercropping crop j and young camellia oleifera plant i. The sensitivity coefficient is determined by fitting the influence degree of different height differences on light attenuation through historical data, and the light shading weight factor is empirically adjusted according to the light competition characteristics of intercropping crop species to young camellia oleifera plant, and the minimum constant is set to 1e-6 to avoid zero denominator.

[0039] Specifically, the canopy density index and the unit leaf area index reflect the overall shading ability of intercropping crop j, and the product value and the reciprocal of the square of the horizontal distance constitute the basic shading intensity term, which embodies the square inverse decay of shading effect with the increase of distance. The height difference is adjusted by an exponential function to adjust the shading intensity. When intercropping crop j is higher than young camellia oleifera plant i, is a positive value, and the exponential term decreases with the increase of height difference, indicating that the increase of height difference can reduce the shading effect; when young camellia oleifera plant i is higher than intercropping crop j, is negative, the exponential term increases to enhance the shielding effect. The sensitivity coefficient controls the degree of non-linear influence of height difference on shielding strength, for example, when = 0.1, the shielding strength decreases by about 9.5% for every 1 meter increase in height difference. The light shading weight factor is set according to the type of intercropping crops, for example, set to = 0.8 for high-stem crops, and set to = 0.3 for low-stem herbaceous crops. By inputting the dynamically calculated light shading coefficient into the graph neural network, the influence of the light shading effect of intercropping crops on the growth of camellia seedlings can be accurately separated, thereby accurately identifying growth abnormalities caused by resource competition.

[0040] Specifically, in the water competition function, the root water absorption capacity and the coverage radius determine the water absorption efficiency of intercropping crops, and the horizontal distance square term reflects the influence of spatial distribution on water competition; in the allelopathic inhibition diffusion function, the allelochemical concentration coefficient and the exponential decay term jointly determine the change law of inhibition strength with distance; in the light shading exponential function, the canopy density and the leaf area index represent the light shading ability of intercropping crops, and the horizontal distance and the height difference jointly modify the shielding strength. The dynamic edge weight values output by the three functions are input into the graph neural network, and through the superposition calculation of multi-source competition relationships, the theoretical height value after removing the intercropping interference can be accurately predicted. When the theoretical height value is lower than the preset threshold, the intercropping interference factor can be excluded, and the growth abnormality of the camellia seedling can be directly determined.

[0041] Through the above technical solutions, the resource competition relationship between the camellia seedling and the intercropping crops can be quantitatively described, so that the influence of intercropping on the growth of the camellia seedling can be more accurately evaluated. Since water competition, allelopathic inhibition, and light shading are considered as multiple ecological factors, this scheme can comprehensively reflect the growth status of the camellia seedling under the intercropping environment. This ecological modeling-based method can improve the accuracy and reliability of camellia seedling growth monitoring, and is helpful for timely discovering and solving growth abnormality problems, thereby improving the cultivation and management quality of camellia seedlings.

[0042] In some of the above schemes of the present application, the interaction between the camellia seedling and the intercropping crops is reflected through the edge relationship in the graph data structure, but in actual scenarios, when the spatial distance between the camellia seedling and the intercropping crops is too large, there may be no substantial resource competition relationship between them. At this time, if the corresponding edge relationship is retained, it will cause invalid calculation when the graph neural network model processes the graph data, thereby affecting the prediction accuracy of the theoretical height value.

[0043] The application further proposes that before extracting the corresponding ecological modeling function, the following steps are included: calculating the spatial distance between each camellia sapling and the adjacent intercropping crop according to the positions of the camellia saplings and the intercropping crops in the image data in the target camellia young forest area; determining whether the spatial distance is greater than a preset distance threshold, and if so, deleting the edge relationship between the corresponding camellia sapling node and the intercropping crop node in the atlas data structure.

[0044] Specifically, after obtaining the image data, the pixel coordinates of the camellia saplings and the intercropping crops are identified by a target detection algorithm, and the horizontal distance between the two is calculated after converting to actual spatial coordinates. The preset distance threshold is dynamically adjusted according to the type of intercropping crop, for example, for intercropping crops with well-developed root systems, the threshold is set to 1.5 times the root coverage radius; for intercropping crops with dense canopy, the threshold is set to 2 times the canopy projection radius. When the calculated spatial distance exceeds the threshold, it is determined that there is no resource competition relationship between the two, and the corresponding edge relationship is automatically removed. This process effectively eliminates the redundant connections in the atlas data, reduces the processing of invalid relationships by the graph neural network model when calculating the dynamic edge weight, improves the model operation efficiency, and avoids the prediction bias caused by retaining invalid edges.

[0045] Through the above technical solutions, the application can effectively reduce the computational complexity and improve the system running efficiency. By setting a reasonable distance threshold, intercropping crops that have little effect on the growth of camellia saplings can be filtered out, making the ecological modeling more focused on key influencing factors. At the same time, this edge relationship screening method based on spatial distance also improves the accuracy of the atlas data structure, providing a more reliable data foundation for subsequent ecological modeling and growth prediction.

[0046] In some of the above schemes of the application, since different camellia varieties have different growth rate characteristics in the young forest stage, if the preset acceptable height value is a fixed value, it may not accurately match the expected growth curve of a specific variety, thereby causing a risk of misjudgment.

[0047] The application further proposes that the preset acceptable height value is dynamically adjusted, specifically including: obtaining the camellia variety identifier of the target camellia young forest area; according to the camellia variety identifier, extracting the corresponding acceptable height value from the preset variety-time-growth threshold mapping table as the preset acceptable height value.

[0048] Among them, the camellia variety identifier is automatically identified through the leaf shape characteristics in the image data or the geographic coding information of the planting area. The variety-time-growth threshold mapping table stores the theoretical height range of different varieties in different growth cycles, and this table is constructed based on historical observation data and contains the correlation between genotype and phenotypic growth parameters. The dynamic adjustment mechanism is realized by real-time matching of the current monitoring time point and the growth stage threshold corresponding to the variety.

[0049] Specifically, in the initialization phase, the camellia variety is identified by the leaf image classification model, and the corresponding variety identification code is generated. The identification code is used as an index key to retrieve the acceptable height value matching the current monitoring date from the mapping table. For example, when monitoring variety A in early July, the system automatically extracts the minimum height threshold of 42 centimeters defined for this variety during the standard growth period in July. When the theoretical height value output by the graph neural network is lower than this threshold, an abnormal alarm is triggered. This scheme eliminates false positives caused by genetic characteristic differences by introducing dynamic thresholds for varieties, making the growth anomaly detection standard strictly aligned with the biological characteristics of the variety.

[0050] Through the above technical solutions, the present application can dynamically adjust the growth evaluation standard according to the growth characteristics of different camellia varieties, avoiding false positives that may occur when using a uniform standard. At the same time, the influence of time and environmental factors on camellia growth is considered, making the anomaly detection more accurate. This flexible evaluation mechanism helps to discover camellia seedlings with growth abnormalities in a timely manner, providing a basis for subsequent precise maintenance, thereby improving the overall growth quality of camellia seedlings.

[0051] Exemplary system

[0052] Figure 2 The deep learning-based camellia seedling growth monitoring system according to an embodiment of the present application is illustrated, which includes a data acquisition module that acquires image data and ground environmental sensing data covering the target camellia seedling area; a graph generation module that generates a graph data structure corresponding to the target camellia seedling area and the current height value of each camellia seedling based on the image data and environmental sensing data, the graph data structure including camellia seedling nodes, intercropping crop nodes, and corresponding edge relationships, and the camellia seedling nodes being bound to the current height value; a function extraction module that extracts corresponding ecological modeling functions from a pre-set relationship-function mapping table based on the edge relationships; an edge weight calculation module that calculates dynamic edge weight values reflecting the resource competition relationship between each camellia seedling and adjacent intercropping crops based on the ecological modeling functions and environmental sensing data; a graph update module that updates the weights of each edge in the graph data structure based on the dynamic edge weight values; a growth prediction module that inputs the updated graph structure data into a pre-trained graph neural network model to output the theoretical height value of each camellia seedling in the target camellia seedling area after removing the influence of intercropping; and a decision module that determines whether the theoretical height value is less than a pre-set acceptable height value, and if so, determines that the corresponding camellia seedling has a growth anomaly.

[0053] In one example, the ecological modeling functions in the function extraction module include at least one of the following functions: a water competition function for representing the relationship between soil water absorption of camellia sinensis seedlings and intercropping crops; an allelopathic inhibition diffusion function for depicting the influence of allelochemicals released by intercropping crops on camellia sinensis seedlings; and a light shading index function for quantifying the light attenuation intensity caused by the shading of intercropping crops to camellia sinensis seedlings.

[0054] In one example, the water competition function extracted by the function extraction module is specifically:

[0055] ;

[0056] wherein, represents the water competition coefficient between camellia sinensis seedling i and intercropping crop j, represents the root water absorption capacity per unit time of crop j, represents the root coverage radius of intercropping crop j, represents the horizontal distance between camellia sinensis seedling i and crop j, is a preset minimum value set to avoid division by zero error.

[0057] In one example, the allelopathic inhibition diffusion function extracted by the function extraction module is specifically:

[0058] ;

[0059] wherein, represents the allelopathic inhibition intensity of intercropping crop j on camellia sinensis seedling i, represents the concentration coefficient of allelochemicals released by intercropping crop j per unit time, represents the horizontal distance between camellia sinensis seedling i and crop j, represents the diffusion attenuation coefficient of allelochemicals released by intercropping crop j in the soil.

[0060] In one example, the light shading index function extracted by the function extraction module is specifically:

[0061] ;

[0062] wherein, represents the light shading influence intensity of intercropping crop j on camellia sinensis seedling i, represents the canopy density index of intercropping crop j, represents the unit leaf area index of intercropping crop j, represents the horizontal distance between camellia sinensis seedling i and crop j, represents the current height difference between intercropping crop j and camellia sinensis seedling i, represents the sensitivity coefficient of vertical height difference to shading degree, representing an illumination occlusion weight factor, a very small constant set to avoid division by zero errors.

[0063] In one example, the function extraction module further includes, before extracting the corresponding ecological modeling function, calculating spatial distances between each tea sapling and the adjacent intercropped crops according to positions of the tea saplings and the intercropped crops in the image data; and determining whether the spatial distances are greater than a preset distance threshold, and if so, deleting an edge relationship between a corresponding tea sapling node and an intercropped crop node in the graph data structure.

[0064] In one example, the preset acceptable height value in the decision module is dynamically adjusted, and specifically includes: obtaining a tea variety identifier of the target tea sapling area; and according to the tea variety identifier, extracting a corresponding acceptable height value from a preset variety-time-growth threshold mapping table as the preset acceptable height value.

[0065] Example electronic device

[0066] Figure 3 An electronic device according to embodiments of the application is illustrated. The electronic device can be the mobile device itself, or a standalone device that can communicate with the mobile device to receive input signals collected therefrom and send selected target driving behaviors thereto.

[0067] Figure 3 A block diagram of an electronic device according to embodiments of the application is illustrated.

[0068] As Figure 3 illustrated, the electronic device includes one or more processors and a memory.

[0069] The processor can be a central processing unit (CPU) or other form of processing unit that has data processing and / or instruction execution capabilities, and can control other components in the electronic device to perform desired functions.

[0070] The memory can include one or more computer program products that can include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory, for example, can include random access memory (RAM), cache, and / or the like. Non-volatile memory, for example, can include read only memory (ROM), hard disk, flash memory, and / or the like. One or more computer program instructions can be stored on the computer-readable storage media, and the processor can execute the program instructions to implement the driving behavior decision method of various embodiments of the application described above and / or other desired functions.

[0071] In one example, the electronic device can further include an input device and an output device, which components are interconnected through a bus system and / or other forms of connection mechanism (not shown).

[0072] Of course, in order to simplify, Figure 3 Only some of the components in the electronic device related to the present application are shown in the figure, and components such as buses, input / output interfaces, etc. are omitted. In addition, according to specific application circumstances, the electronic device can also include any other appropriate components.

[0073] Exemplary computer readable medium

[0074] Embodiments of the present application can also be a computer readable storage medium, which stores computer program instructions, and the computer program instructions, when executed by a processor, cause the processor to perform the steps in the driving behavior decision method according to various embodiments of the present application described in the above "Exemplary Method" section of the specification.

[0075] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium may, for example, include but is not limited to an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device or instrument, or any combination of the above. More specific examples (non-exhaustive list) of readable storage medium include: electrical connection with one or more conductive wires, portable disk, hard disk, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.

[0076] The basic principles of the present application are described above in combination with specific embodiments, but it should be noted that the advantages, advantages, effects, etc. mentioned in the present application are only examples and are not limiting, and these advantages, advantages, effects, etc. cannot be considered as the must-have of each embodiment of the present application. In addition, the above specific details are only for the purpose of example and for the purpose of understanding, and are not limiting, and the above details do not limit the present application to the must-use of the above specific details to realize.

[0077] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0078] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0079] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0080] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A deep learning-based method for monitoring the growth of young Camellia oleifera forests, characterized in that, include: Acquire image data and ground environment sensor data covering the target Camellia oleifera sapling forest area; Based on the image data and the environmental sensing data, a map data structure corresponding to the target Camellia oleifera sapling area and the current height value of each Camellia oleifera sapling are generated. The map data structure includes Camellia oleifera sapling nodes, intercropping crop nodes and corresponding edge relationships. The Camellia oleifera sapling nodes are bound to the current height value. Based on the edge relationship, the corresponding ecological modeling function is extracted from the preset relation-function mapping table; the ecological modeling function includes at least one of the following functions: a water competition function to represent the soil water absorption relationship between Camellia oleifera seedlings and intercropped crops; an allelopathic inhibition diffusion function to characterize the effect of allelopathic substances released by intercropped crops on Camellia oleifera seedlings; and a light shading index function to quantify the light attenuation intensity caused by intercropped crops shading Camellia oleifera seedlings. The dynamic edge weights, which reflect the resource competition relationship between each camellia oleifera seedling and adjacent intercropped crops, are calculated based on the ecological modeling function and the environmental sensing data. The weights of each edge in the graph data structure are updated according to the dynamic edge weights; The updated graph data structure is input into the pre-trained graph neural network model, which outputs the theoretical height of each Camellia oleifera seedling in the target Camellia oleifera seedling area at the current time point after removing the influence of intercropping interference. Determine whether the theoretical height value is less than the preset acceptable height value. If so, determine that the corresponding camellia oleifera seedling is growing abnormally.

2. The method for monitoring the growth of young Camellia oleifera forests based on deep learning according to claim 1, characterized in that, The water competition function is specifically as follows: ； in, This represents the water competition coefficient between young camellia oleifera plant i and intercropped crop j. This represents the root system's water absorption capacity per unit time for crop j. This represents the root coverage radius of the intercropped crop j. This represents the horizontal distance between camellia oleifera seedling i and crop j. The preset minimum value is set to avoid division by zero errors.

3. The method for monitoring the growth of young Camellia oleifera forests based on deep learning according to claim 1, characterized in that, The allelopathic suppression diffusion function is specifically: ； in, This indicates the allelopathic inhibition intensity of intercropped crop j on young camellia oleifera plantlets i. This represents the concentration coefficient of allelochemicals released by intercropped crop j per unit time. This represents the horizontal distance between camellia oleifera seedling i and crop j. This represents the diffusion attenuation coefficient of allelochemicals released by intercropped crop j in the soil.

4. The method for monitoring the growth of young Camellia oleifera forests based on deep learning according to claim 1, characterized in that, The light shading exponential function is specifically as follows: ； in, This indicates the intensity of the light shading effect of intercropping crop j on young camellia oleifera plants i. This represents the canopy density index of intercropped crop j. This represents the leaf area index per unit of intercropped crop j. This represents the horizontal distance between camellia oleifera seedling i and crop j. This represents the current height difference between the intercropped crop j and the young camellia oleifera plant i. This represents the sensitivity coefficient of vertical height difference to the degree of shading. This represents the light shading weight factor. A very small constant set to avoid division by zero errors.

5. The method for monitoring the growth of young Camellia oleifera forests based on deep learning according to claim 1, characterized in that, Before extracting the corresponding ecological modeling function, the following also applies: Based on the positions of each Camellia oleifera sapling and intercropped crop in the target Camellia oleifera sapling forest area in the image data, calculate the spatial distance between each Camellia oleifera sapling and the adjacent intercropped crop. Determine whether the spatial distance is greater than a preset distance threshold. If so, delete the edge relationship between the corresponding camellia oleifera seedling node and the intercropping node in the graph data structure.

6. The method for monitoring the growth of young Camellia oleifera forests based on deep learning according to claim 1, characterized in that, The preset acceptable height value is dynamically adjusted, specifically including: Obtain identification of Camellia oleifera varieties in the target Camellia oleifera sapling area; Based on the camellia oleifera variety identifier, the corresponding acceptable height value is extracted from the preset variety-time-growth threshold mapping table as the preset acceptable height value.

7. A deep learning-based growth monitoring system for young Camellia oleifera forests, characterized in that, include: The data acquisition module acquires image data and ground environment sensor data covering the target camellia oleifera sapling forest area; The map generation module generates a map data structure corresponding to the target camellia oleifera sapling area and the current height value of each camellia oleifera sapling based on the image data and the environmental sensing data. The map data structure includes camellia oleifera sapling nodes, intercropping crop nodes and corresponding edge relationships. The camellia oleifera sapling nodes are bound to the current height value. The function extraction module extracts the corresponding ecological modeling function from a preset relation-function mapping table based on the edge relationship. The ecological modeling function includes at least one of the following functions: a water competition function to represent the soil water absorption relationship between Camellia oleifera seedlings and intercropped crops; an allelopathic inhibition diffusion function to characterize the effect of allelopathic substances released by intercropped crops on Camellia oleifera seedlings; and a light shading index function to quantify the light attenuation intensity caused by intercropped crops shading Camellia oleifera seedlings. The edge weight calculation module calculates dynamic edge weight values ​​based on the ecological modeling function and the environmental sensing data to reflect the resource competition relationship between each camellia oleifera seedling and adjacent intercropped crops. The graph update module updates the weights of each edge in the graph data structure according to the dynamic edge weights; The growth prediction module inputs the updated map data structure into the pre-trained graph neural network model and outputs the theoretical height values ​​of each Camellia oleifera seedling in the target Camellia oleifera seedling area after removing the influence of intercropping interference. The decision module determines whether the theoretical height value is less than the preset acceptable height value. If so, it determines that the corresponding camellia oleifera seedling is growing abnormally.

8. An electronic device comprising a memory and a processor, characterized in that: The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the method as described in any one of claims 1 to 7.

9. A computer storage medium storing computer-executable instructions thereon, characterized in that: When the computer-executable instructions are executed by a processor, they implement the steps of the method as described in any one of claims 1 to 7.

Images