An intelligent garden monitoring system and method based on image recognition

Abstract

The application discloses an intelligent garden monitoring system and method based on image recognition, and belongs to the technical field of garden monitoring. The method comprises the following steps: a perception end collects real-time data of a garden, and transmits the garden collected data to a cloud platform after preprocessing; the cloud platform divides the garden into different zones based on various garden analysis requirements and the garden collected data, obtains functional zones corresponding to the various garden analysis requirements and zone correlation information, and performs real-time update evaluation on each functional zone; the functional zones and the analysis correlation information that need to be updated are updated correspondingly; the garden collected data of the corresponding functional zones are analyzed through a preset functional analysis model, a demand analysis result is obtained, the demand analysis results corresponding to the various garden analysis requirements are summarized as a monitoring analysis result, and the monitoring analysis result is sent to the perception end.

Description

Technical Field

[0001] This invention belongs to the field of garden monitoring technology, specifically a smart garden monitoring system and method based on image recognition. Background Technology

[0002] With the rapid development of smart garden construction, image recognition-based intelligent monitoring systems have gradually replaced traditional manual inspections and are widely used in garden safety protection, fire early warning, and vegetation maintenance. Currently, most mainstream monitoring systems in the industry adopt fixed, general-purpose AI visual analysis models. Their algorithm parameters, recognition logic, and monitoring dimensions are fixed, employing a uniform "one-size-fits-all" recognition mode across the entire garden. This lack of scene-adaptive adjustment capabilities makes them unable to adapt to the dynamic changes in garden ecological scenes, highlighting their technical shortcomings. Garden vegetation exhibits significant temporal dynamic changes, continuously altering its appearance with seasonal growth, decay, and pest outbreaks. Furthermore, outdoor gardens experience dynamic interference from light and shadow, rain and snow, and vegetation obstruction, causing real-time fluctuations in scene characteristics. In addition, different garden areas have significantly different risk types, control priorities, and vegetation growth characteristics, resulting in varying requirements for image recognition monitoring dimensions and accuracy thresholds. Traditional fixed models cannot dynamically adjust recognition strategies to follow vegetation growth and scene changes, leading to unstable recognition accuracy at different vegetation growth stages and problems such as misjudgment of vegetation status, missed fire detection, and false intrusion alarms.

[0003] In order to solve the above problems, the present invention provides a smart garden monitoring system and method based on image recognition. Summary of the Invention

[0004] To address the problems of the aforementioned solutions, this invention provides a smart garden monitoring system and method based on image recognition.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A smart garden monitoring system based on image recognition, comprising a cloud platform and sensing terminals;

[0007] A smart garden monitoring system based on image recognition, comprising a cloud platform and sensing terminals;

[0008] Furthermore, the cloud platform and the sensing terminal are connected via communication.

[0009] The sensing terminal is used to collect real-time data from the garden, preprocess the collected data, and transmit it to the cloud platform; it also receives monitoring and analysis results from the cloud platform and performs corresponding processing based on these results.

[0010] The cloud platform includes a model library, a garden zoning module, and a modular analysis module;

[0011] The model library is used to store various functional analysis models.

[0012] Furthermore, the various functional analysis models stored in the model library are classified according to the corresponding analysis sub-units.

[0013] The garden zoning module identifies various garden analysis needs, acquires garden data, and zons the garden based on these needs and data to obtain the functional zones and their associated information. It also performs real-time updates and evaluations on each functional zone and updates the necessary functional zones and their associated information accordingly.

[0014] Furthermore, based on various garden analysis needs and garden data collection, gardens are divided into zones, including:

[0015] Identify all potential garden hazards for each garden analysis requirement, obtain garden information, and filter all potential garden hazards for each garden analysis requirement based on the garden information to obtain the various potential garden hazards that the garden has for the garden analysis requirement.

[0016] Identify the functional analysis models corresponding to various garden analysis needs, conduct simulation analysis based on each functional analysis model, garden collected data, and potential garden hazards, obtain the application areas corresponding to each functional analysis model, and summarize the application areas corresponding to each functional analysis model into a functional partition.

[0017] Furthermore, based on the garden information, a comprehensive screening of potential garden hazards for each garden analysis requirement is performed, including:

[0018] A hazard screening model is established, and the expression of the hazard screening model is as follows:

[0019]

[0020] In the formula: (s) i (YU) represents the input data, s i This indicates the corresponding potential garden hazards in the total garden hazard list for the garden analysis requirements, where i is the subscript, i = 1, 2, ..., n, and n is the number of potential garden hazards in the total garden hazard list; YU represents garden information; s i Abnormal data indicates that, based on the analysis of the garden information, the potential garden hazard will not occur or the probability of occurrence is lower than the preset value; the output data is the hazard screening value D. ang (s) i (YU), the hazard screening value is 1 or 0;

[0021] The corresponding potential garden hazards and garden information from the full set of garden hazards corresponding to the garden analysis needs are combined into input data and input into the hazard screening model for analysis to obtain the hazard screening value of the potential garden hazards;

[0022] Potential garden hazards with a hazard screening value of 1 are eliminated.

[0023] Furthermore, simulation analysis is conducted based on various functional analysis models, garden data collection, and potential garden hazards, including:

[0024] Based on the collected garden data and potential garden hazards, various simulation data are set up, and the garden area is divided into several unit areas. The simulation data are analyzed by various functional analysis models for garden analysis needs, and the hazard identification time and hazard identification accuracy of the functional analysis models under the background of potential garden hazards in the corresponding unit areas are obtained.

[0025] Calculate the priority value of each functional analysis model in the corresponding unit region according to the priority formula; the priority formula is:

[0026]

[0027] In the formula: Priority value; The weighting coefficients are for the corresponding potential garden hazards. =1; j represents the corresponding potential garden hazard background, j=1, 2, ..., m, where m is the number of potential garden hazard backgrounds in this unit area; To assess the accuracy of the functional analysis model in identifying potential garden hazards within this unit area; The time required for the functional analysis model to identify potential garden hazards in this unit area; The minimum time required for hazard identification; The maximum allowable time for hazard identification; To prevent extremely small positive numbers with a denominator of zero;

[0028] Assign the functional analysis model with the highest priority value to the unit region; merge unit regions that are adjacent in position and have the same corresponding functional analysis model to obtain a new unit region, and so on, until there are no unit regions that meet the merging requirements, and mark the remaining unit regions as application regions.

[0029] Furthermore, real-time updates and evaluations are conducted on each functional area, including:

[0030] Obtain benchmark garden data corresponding to each functional zone, wherein the benchmark garden data is the garden data used when determining the functional zone; compare the real-time garden data with the benchmark garden data to determine whether there are areas with excessive differences;

[0031] When the analysis shows no areas with excessive differences, the assessment does not require updating the functional partitions;

[0032] When analyzing areas with excessive differences, simulation analysis is performed based on various functional analysis models, garden data collected from areas with excessive differences, and potential garden hazards to obtain the application area corresponding to the area with excessive differences and the functional analysis model adapted to the application area. When the area with excessive differences corresponds to a functional analysis model that does not correspond to the functional partition, the evaluation requires updating the functional partition; otherwise, the evaluation does not require updating the functional partition.

[0033] The modular analysis module configures corresponding analysis sub-units according to the garden analysis requirements, identifies each functional zone, and the analysis sub-unit loads the functional analysis model corresponding to each functional zone of the garden analysis requirements from the model library.

[0034] The system acquires garden data, and the analysis subunit analyzes the garden data of the corresponding functional areas through the functional analysis model to obtain the demand analysis results. The demand analysis results corresponding to each garden analysis requirement are summarized into monitoring analysis results, and the monitoring analysis results are sent to the sensing end.

[0035] Furthermore, each analysis subunit is a modular unit that can be dynamically adjusted according to the needs of landscape analysis.

[0036] Furthermore, landscape analysis needs include fire risk analysis needs, landscape safety protection analysis needs, and plant health status analysis needs.

[0037] A smart garden monitoring method based on image recognition, the method includes:

[0038] The sensing device collects real-time data from the garden, and transmits the collected data to the cloud platform after preprocessing.

[0039] The cloud platform divides the garden into zones based on various garden analysis needs and garden collection data, obtaining the functional zones corresponding to each garden analysis need and the zone association information; it performs real-time update and evaluation of each functional zone, and updates the functional zones and analysis association information that need to be updated accordingly;

[0040] The garden data collected in the corresponding functional areas is analyzed by a preset functional analysis model to obtain the demand analysis results. The demand analysis results corresponding to each garden analysis requirement are summarized into monitoring analysis results and sent to the sensing end.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] This invention relies on a dynamically adaptable image recognition architecture, breaking away from the industry limitations of traditional smart garden monitoring systems that use fixed, general AI models and operate uniformly across the entire garden. It no longer depends on fixed algorithm parameters and recognition logic, but can deeply adapt to the natural dynamic attributes of garden ecological scenes. It can perform real-time perception and strategy adjustment for changes in the appearance of vegetation due to seasonal growth and decay, outbreaks of pests and diseases, and complex variables such as light and shadow fluctuations, rain and snow interference, and vegetation shading in outdoor scenes. It fundamentally avoids the problem of large fluctuations in recognition accuracy at different growth stages in traditional solutions. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a block diagram illustrating the principle of the present invention. Detailed Implementation

[0045] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] like Figure 1 As shown, a smart garden monitoring system based on image recognition includes a cloud platform and a sensing terminal;

[0047] The cloud platform and the sensing terminal communicate with each other.

[0048] The sensing terminal is deployed in the garden to collect real-time data from the garden, preprocess the data and transmit it to the cloud platform, receive the monitoring and analysis results sent by the cloud platform, and perform corresponding processing based on the monitoring and analysis results, such as early warning.

[0049] For example, a system comprised of high-definition night vision cameras, panoramic gimbal cameras, and fixed-point zoom cameras at various monitoring points within the park enables real-time, all-weather image data acquisition across the entire park area. This can be supplemented dynamically by combining drones and other equipment. Simultaneously, the sensing end incorporates on-device data preprocessing logic to perform real-time cleaning, cropping, noise reduction, lighting correction, invalid frame removal, and resolution standardization of the raw image data. This filters out image noise caused by environmental interference from outdoor park scenes such as rain, snow, strong light, and camera shake, and eliminates blank, blurry, and invalid monitoring images. The preprocessed standardized image data is stably transmitted to the cloud platform via an encrypted network, significantly reducing the cloud data processing load and improving overall recognition and analysis efficiency and real-time performance.

[0050] The cloud platform includes a model library, a garden zoning module, and a modular analysis module;

[0051] The model library is used to store various functional analysis models.

[0052] In one embodiment, the functional analysis models are classified according to the corresponding analysis sub-units.

[0053] For example, various functional analysis models used for security protection are adapted to security protection analysis sub-units. They are classified according to the adapted analysis sub-units. Functional analysis models adapted to the same analysis sub-unit are classified into one category. Generally, when setting up or establishing a functional analysis model based on the analysis sub-unit, corresponding association tags are made to facilitate subsequent calls.

[0054] The garden zoning module acquires garden data and identifies various garden analysis needs, such as common fire risk analysis needs, garden safety protection analysis needs, and plant health status analysis needs, corresponding to the various analysis sub-units in the analysis module. Based on each garden analysis need and garden data, the module zons the garden to obtain the functional zones corresponding to each garden analysis need and the zone association information. The zone association information mainly includes the garden analysis needs associated with the functional zone, the functional analysis model, and other information. The module also updates and evaluates each functional zone, updating the functional zones and analysis association information for those with updated needs.

[0055] In one embodiment, garden zoning is performed based on various garden analysis needs and garden data collection, including:

[0056] Based on relevant historical data, a full range of potential garden hazards for each garden analysis requirement are identified, and garden information is obtained. This garden information is mainly used to analyze whether the garden has potential garden hazards, which mainly includes relevant information for assessing whether it has potential garden hazards, such as fire hazards. Information on whether the garden has fire hazards is collected, which can be combined with the manifestation characteristics and causes of the full range of garden hazards. Based on the garden information, the full range of garden hazards for each garden analysis requirement are filtered to obtain the various potential garden hazards that the garden has for the garden analysis requirement.

[0057] Identify the functional analysis models corresponding to various garden analysis needs, conduct simulation analysis based on each functional analysis model, garden collected data, and potential garden hazards, obtain the application areas corresponding to each functional analysis model, and uniformly mark each application area corresponding to the functional analysis model as a functional partition, that is, a functional analysis model with an application area, and one functional analysis model corresponds to one functional partition.

[0058] In one embodiment, the full range of potential garden hazards for various garden analysis needs are screened based on garden information. This screening can be done using existing methods, such as intelligent screening based on intelligent models built using machine learning; or it can be done manually.

[0059] In one embodiment, based on garden information, a comprehensive screening of potential garden hazards for various garden analysis needs is performed, including:

[0060] A hazard screening model is established, and the expression of the hazard screening model is as follows:

[0061]

[0062] In the formula: (s) i (YU) represents the input data, s i This indicates the corresponding potential garden hazards in the total garden hazard list for the garden analysis requirements, where i is the subscript, i = 1, 2, ..., n, and n is the number of potential garden hazards in the total garden hazard list; YU represents garden information; s i Abnormal data indicates that, based on the analysis of the garden information, the potential garden hazard will not occur in the garden, or the probability of occurrence is lower than the preset value; the output data is the hazard screening value D. ang (s) i (YU), the hazard screening value is 1 or 0; training is performed based on the corresponding historical data, and the training set includes input data and output data;

[0063] The corresponding potential garden hazards and garden information from the full set of garden hazards corresponding to the garden analysis needs are combined into input data and input into the hazard screening model for analysis to obtain the hazard screening value of the potential garden hazards;

[0064] Potential garden hazards with a hazard screening value of 1 are eliminated.

[0065] In one embodiment, simulation analysis is performed based on various functional analysis models, garden data collection, and potential garden hazards, including:

[0066] Various simulation data are set up based on garden collection data and potential garden hazards. For example, assuming a potential garden hazard exists in a certain location, the corresponding garden collection data is adjusted according to the initial appearance of the potential garden hazard at that location to form simulation data. The garden collection data at the corresponding location can be adjusted by combining the historical performance data of the potential garden hazard. For example, normal garden collection data can be gradually adjusted according to the historical development data of the potential garden hazard. Simulation data can also be generated by combining other simulation technologies, such as setting up a digital twin model based on accumulated garden collection data based on digital twin technology, and simulating simulation data under various potential garden hazard conditions based on the digital twin model. The same potential garden hazard at the same location can also be set up with different simulation data, such as differences in severity, development trend, etc., all corresponding to possible actual situations.

[0067] The garden area is divided into several unit areas. Simulation data is analyzed using various functional analysis models tailored to the garden analysis requirements. This yields the hazard identification time and accuracy of the functional analysis models under the background of potential garden hazards in the corresponding unit areas. The potential garden hazard background refers to the actual potential garden hazard situation corresponding to the simulation data. For unit areas without potential garden hazards, the time required to determine that the unit area has no potential garden hazards is the hazard identification time, and the accuracy rate of determining that the unit area has no potential garden hazards is the hazard identification accuracy rate. Multiple simulation analyses are performed in the cloud to determine the hazard identification time and accuracy rate of the functional analysis models under the background of potential garden hazards in the corresponding unit areas. Only the initial analysis is a full-area analysis; subsequent analyses only require determining the functional zones that need to be updated.

[0068] Calculate the priority value of each functional analysis model in the corresponding unit region according to the priority formula; the priority formula is:

[0069]

[0070] In the formula: Priority value; The weighting coefficients are for the corresponding potential garden hazards. =1, which can be used to calculate the weight coefficient based on the probability of occurrence of the potential garden hazard background, or the weight coefficient can be set based on other factors such as loss and risk; j represents the corresponding potential garden hazard background, j=1, 2, ..., m, where m is the number of potential garden hazard backgrounds in the unit area (the number of potential garden hazard backgrounds for simulation). To assess the accuracy of the functional analysis model in identifying potential garden hazards in this unit area; The time required for the functional analysis model to identify potential garden hazards in this unit area; The minimum time for identifying potential hazards is determined by statistical analysis based on relevant historical data. The maximum allowed time for hazard identification can be set according to management needs or adjusted according to user requirements. To prevent extremely small positive numbers with a denominator of zero;

[0071] Assign the functional analysis model with the highest priority value to the unit area; merge adjacent unit areas with the same corresponding functional analysis model to obtain a new unit area, and so on, until there are no unit areas that meet the merging requirements (adjacent and with the same corresponding functional analysis model), and mark the remaining unit areas as application areas.

[0072] In one embodiment, the garden area is divided into several unit areas, and each unit area is divided according to a preset grid size, such as 1m×1m. The grid size can be set according to the minimum analysis area of ​​the functional analysis model.

[0073] In one embodiment, the garden area is divided into several unit areas. The unit areas can be divided based on clustering. First, the minimum unit area is preset, and then the garden data is analyzed based on various clustering algorithms. Clustering is performed according to location attributes (plant species, image features, etc.), such as hierarchical clustering algorithm, machine learning clustering algorithm, etc.

[0074] In one embodiment, real-time evaluation of each functional partition includes:

[0075] Obtain the baseline garden data for each functional zone, i.e., the garden data used when dividing the functional zone; compare the real-time garden data with the baseline garden data to determine if there are any areas with excessive differences.

[0076] When the analysis shows no areas with excessive differences, the assessment does not require updating the functional partitions;

[0077] When analyzing areas with excessive differences, simulation analysis is performed based on various functional analysis models, garden data collected from these areas, and potential garden hazards. This yields the application areas corresponding to the areas with excessive differences and the functional analysis models that fit those areas. If an area with excessive differences corresponds to a functional analysis model that does not belong to the functional area, the evaluation requires updating the functional area. Subsequently, the functional area is adjusted based on the application area and the fitted functional analysis model, and adjacent application areas that fit the same functional analysis model are merged. Conversely, if no such model is found, the evaluation does not require updating the functional area.

[0078] In one embodiment, real-time garden data is compared with baseline garden data to determine whether there are areas with excessive differences. This can be done by identifying areas with excessive differences based on the degree of difference in image features, such as areas with a difference greater than a preset value; or by using intelligent models built based on machine learning, deep learning algorithms, etc., to intelligently identify areas with excessive differences.

[0079] In one embodiment, real-time garden data is compared with baseline garden data to determine whether there are areas with excessive discrepancies, including:

[0080] The system pre-defines various influencing factors that have a monitoring impact on different garden analysis needs. These factors can be combined with those that influence the identification of potential garden hazards to determine the influencing factors, such as shading factors (plant growth shading, snow shading, etc.), meteorological factors, and seasonal factors. Based on each influencing factor, the system performs feature identification on the garden data collected and the baseline garden data collected to obtain the difference in influence characteristics corresponding to each influencing factor. The system then assesses whether any single factor exceeds the standard based on the difference in influence characteristics of each influencing factor. If a single factor exceeds the standard, it is considered a difference exceeding the standard, and the exceeding area is identified and marked as a difference exceeding the standard area.

[0081] A comprehensive analysis can also be performed, normalizing the various differences and influencing characteristics, and calculating based on a preset comprehensive formula, such as a weighted calculation formula. When the comprehensive value is greater than the preset value, it is considered that the difference exceeds the standard.

[0082] For example, the difference characteristics of each influencing item are normalized to obtain the difference values ​​of each individual item. k represents the corresponding influencing item, k = 1, 2, ..., v, where v is the number of influencing items;

[0083] Calculate the comprehensive value HP using the comprehensive formula:

[0084]

[0085] In the formula: These are the weighting coefficients for the corresponding influencing terms.

[0086] The modular analysis module configures corresponding analysis sub-units according to the garden analysis requirements, identifies each functional zone, and the analysis sub-unit loads the functional analysis model corresponding to each functional zone of the garden analysis requirements from the model library.

[0087] The system acquires garden data, and the analysis subunit analyzes the garden data of the corresponding functional areas through the functional analysis model to obtain the demand analysis results. The demand analysis results corresponding to each garden analysis requirement are summarized into monitoring analysis results, and the monitoring analysis results are sent to the sensing end.

[0088] A smart garden monitoring method based on image recognition, the method includes:

[0089] The sensing device collects real-time data from the garden, and transmits the collected data to the cloud platform after preprocessing.

[0090] The cloud platform divides the garden into zones based on various garden analysis needs and garden collection data, obtaining the functional zones corresponding to each garden analysis need and the zone association information; it performs real-time update and evaluation of each functional zone, and updates the functional zones and analysis association information that need to be updated accordingly;

[0091] The garden data collected in the corresponding functional areas is analyzed by a preset functional analysis model to obtain the demand analysis results. The demand analysis results corresponding to each garden analysis requirement are summarized into monitoring analysis results and sent to the sensing end.

[0092] The above formulas are all numerical calculations after removing dimensions. The formulas are obtained by software simulation based on a large amount of data and are closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulation based on a large amount of data.

[0093] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A smart garden monitoring system based on image recognition, characterized in that, This includes cloud platforms and sensing terminals; The sensing terminal is used to collect real-time data from the garden, preprocess the collected data, and transmit it to the cloud platform; it also receives monitoring and analysis results from the cloud platform and performs corresponding processing based on the monitoring and analysis results. The cloud platform includes a model library, a garden zoning module, and a modular analysis module; The model library is used to store various functional analysis models; The garden zoning module identifies various garden analysis needs, acquires garden data, and zons the garden based on these needs to obtain functional zones corresponding to each garden analysis need. It then performs real-time updates and evaluations of each functional zone, updating those that require updates. The modular analysis module configures corresponding analysis sub-units according to the garden analysis requirements, identifies each functional zone, and the analysis sub-unit loads the functional analysis model corresponding to each functional zone of the garden analysis requirements from the model library. The system acquires garden data, and the analysis subunit analyzes the garden data of the corresponding functional areas through the functional analysis model to obtain the demand analysis results. The demand analysis results corresponding to each garden analysis requirement are summarized into monitoring analysis results, and the monitoring analysis results are sent to the sensing end.

2. The intelligent garden monitoring system based on image recognition according to claim 1, characterized in that, The cloud platform and the sensing terminal communicate with each other.

3. The intelligent garden monitoring system based on image recognition according to claim 1, characterized in that, The various functional analysis models stored in the model library are categorized according to the corresponding analysis sub-units.

4. The intelligent garden monitoring system based on image recognition according to claim 1, characterized in that, Based on various garden analysis needs and garden data collection, gardens are divided into zones, including: Identify all potential garden hazards for each garden analysis requirement, obtain garden information, and filter all potential garden hazards for each garden analysis requirement based on the garden information to obtain the various potential garden hazards that the garden has for the garden analysis requirement. Identify the functional analysis models corresponding to various garden analysis needs, conduct simulation analysis based on each functional analysis model, garden collected data, and potential garden hazards, obtain the application areas corresponding to each functional analysis model, and summarize the application areas corresponding to each functional analysis model into a functional partition.

5. The intelligent garden monitoring system based on image recognition according to claim 4, characterized in that, Based on garden information, a comprehensive screening of potential garden hazards is performed for various garden analysis needs, including: A hazard screening model is established, and the expression of the hazard screening model is as follows: In the formula: (s) i (YU) represents the input data, s i This indicates the corresponding potential garden hazards in the total garden hazard list for the garden analysis requirements, where i is the subscript, i = 1, 2, ..., n, and n is the number of potential garden hazards in the total garden hazard list; YU represents garden information; s i Abnormal data indicates that, based on the analysis of the garden information, the potential garden hazard will not occur or the probability of occurrence is lower than the preset value; the output data is the hazard screening value D. ang (s) i (YU), the hazard screening value is 1 or 0; The corresponding potential garden hazards and garden information from the full set of garden hazards corresponding to the garden analysis needs are combined into input data and input into the hazard screening model for analysis to obtain the hazard screening value of the potential garden hazards; Potential garden hazards with a hazard screening value of 1 are eliminated.

6. The intelligent garden monitoring system based on image recognition according to claim 4, characterized in that, Simulation analysis was conducted based on various functional analysis models, garden data collection, and potential garden hazards, including: Based on the collected garden data and potential garden hazards, various simulation data are set up, and the garden area is divided into several unit areas. The simulation data are analyzed by various functional analysis models for garden analysis needs, and the hazard identification time and hazard identification accuracy of the functional analysis models under the background of potential garden hazards in the corresponding unit areas are obtained. Calculate the priority value of each functional analysis model in the corresponding unit region according to the priority formula; the priority formula is: In the formula: Priority value; The weighting coefficients are for the corresponding potential garden hazards. =1; j represents the corresponding potential garden hazard background, j=1, 2, ..., m, where m is the number of potential garden hazard backgrounds in this unit area; To assess the accuracy of the functional analysis model in identifying potential garden hazards within this unit area; The time required for the functional analysis model to identify potential garden hazards in this unit area; The minimum time required for hazard identification; The maximum allowable time for hazard identification; To prevent extremely small positive numbers with a denominator of zero; Assign the functional analysis model with the highest priority value to the unit region; merge adjacent unit regions with the same corresponding functional analysis model to obtain a new unit region, and so on, until there are no unit regions that meet the merging requirements, and mark the remaining unit regions as application regions.

7. The intelligent garden monitoring system based on image recognition according to claim 6, characterized in that, Real-time evaluation and updates are performed on each functional area, including: Obtain benchmark garden data corresponding to each functional zone, wherein the benchmark garden data is the garden data used when determining the functional zone; compare the real-time garden data with the benchmark garden data to determine whether there are areas with excessive differences; When the analysis shows no areas with excessive differences, the assessment does not require updating the functional partitions; When analyzing areas with excessive differences, simulation analysis is performed based on various functional analysis models, garden data collected from areas with excessive differences, and potential garden hazards to obtain the application area corresponding to the area with excessive differences and the functional analysis model adapted to the application area. When the area with excessive differences corresponds to a functional analysis model that does not correspond to the functional partition, the evaluation requires updating the functional partition; otherwise, the evaluation does not require updating the functional partition.

8. The intelligent garden monitoring system based on image recognition according to claim 1, characterized in that, Each analysis subunit is a modular unit that can be dynamically adjusted according to the needs of landscape analysis.

9. A smart garden monitoring system based on image recognition according to claim 1, characterized in that, Landscape analysis needs include fire risk analysis, landscape safety protection analysis, and plant health status analysis.

10. A smart garden monitoring method based on image recognition, characterized in that, The method, applied to an image recognition-based smart garden monitoring system as described in any one of claims 1 to 9, comprises: The sensing device collects real-time data from the garden, and transmits the collected data to the cloud platform after preprocessing. The cloud platform divides the garden into zones based on various garden analysis needs and garden collection data, obtaining the functional zones corresponding to each garden analysis need and the zone association information; it performs real-time update and evaluation of each functional zone, and updates the functional zones and analysis association information that need to be updated accordingly; The garden data collected in the corresponding functional areas is analyzed by a preset functional analysis model to obtain the demand analysis results. The demand analysis results corresponding to each garden analysis requirement are summarized into monitoring analysis results and sent to the sensing end.

Images