A non-supplementary light AI identification monitoring method and system for garden invasive insects

Abstract

The application discloses a kind of no light supplement AI identification monitoring methods and systems for garden invasive insects, belong to the field of intelligent monitoring of garden plant protection, including the monitoring terminal of the layout carrying sticky insect tape and no light supplement low-illumination imaging equipment;In the condition without active light supplement, only rely on environmental natural light or night weak light to carry out all-weather image snapshot;Edge AI identification terminal local operation insect identification model, output the structured data of category, quantity and insect density;Through wireless communication, upload to cloud platform to carry out insect situation statistics and alarm research and judge.The application discards active light supplement to avoid light source trend lead to data distortion, combined with edge computing to realize local real-time identification and unknown species automatic collection, suitable for all-weather automatic monitoring of invasive insects in garden greening, park, forest area and other scenes.

Description

Technical Field

[0001] This invention belongs to the field of intelligent monitoring of garden plant protection, and more specifically, it relates to an AI-based identification and monitoring method and system for invasive insects in gardens without supplemental lighting. Background Technology

[0002] With the rapid advancement of refined and intelligent urban landscaping in my country, the area of ​​urban green spaces continues to expand, and the types of green spaces are becoming increasingly fragmented and diverse. At the same time, the inter-regional transport and natural migration of seedlings have led to frequent outbreaks of invasive insects in urban areas. Quarantine-related invasive insects such as the fall webworm, the sycamore lace bug, and the red imported fire ant are characterized by rapid reproduction, lack of native predators, strong concealment, and wide spread, easily causing large-scale damage to green vegetation and even threatening urban ecological security.

[0003] Garden insect monitoring is a core foundation for pest control. Currently, mainstream monitoring schemes are mainly divided into two categories:

[0004] The first category is the traditional manual monitoring method, which relies on maintenance personnel to regularly patrol the green areas, using equipment such as yellow sticky insect traps, and manually counting the number of insects and recording the occurrence of pests. This method has problems such as low efficiency, strong lag, high labor costs, strong subjectivity of data, and inability to form continuous monitoring data.

[0005] The second category is the traditional intelligent insect monitoring scheme, which uses intelligent insect monitoring lamps with active supplemental lighting or high-definition monitoring cameras to attract photophilic insects through active light sources at night, and then uploads the images to the cloud for AI recognition. However, this scheme has the following insurmountable defects: (1) Active supplemental lighting will attract a large number of photophilic insects, which will seriously change the activity patterns of insects in the natural state, resulting in distorted monitoring data and large errors in insect population density reporting; (2) It relies on centralized computing power in the cloud, which is prone to data loss and recognition delay in areas with weak networks, resulting in insufficient real-time performance; (3) The equipment has poor protection, and long-term outdoor deployment is prone to failure due to rodents and insects gnawing, dampness, etc., resulting in high maintenance costs; (4) The data dimension is single, usually focusing only on pests, and cannot simultaneously distinguish between invasive insects, native pests and beneficial insects, making it difficult to support ecological surveys and long-term scientific research; (5) The construction of power supply wiring is highly destructive, relying on the mains power and wired network, and the cost of deployment in scattered green areas is extremely high, making it difficult to achieve full-area grid monitoring. Summary of the Invention

[0006] The main objective of this invention is to provide a method and system for AI-based identification and monitoring of invasive insects in gardens without supplemental lighting, enabling interference-free, high-precision, all-weather, and fully automated monitoring of invasive insects.

[0007] According to a first aspect of the present invention, an AI-based identification and monitoring method for invasive garden insects without supplemental lighting is provided, comprising the following steps:

[0008] S1: In the garden monitoring scenario, a monitoring terminal is deployed. The monitoring terminal is equipped with sticky insect tape as an insect trapping carrier and is equipped with a low-light imaging device without supplemental lighting.

[0009] S2: Using the aforementioned low-light imaging device without supplementary lighting, under conditions without active light source supplementary lighting, relying only on ambient natural light or weak urban night light, the insect targets on the sticky tape are captured in real time and / or recorded in dynamic video to obtain insect image data.

[0010] S3: The edge AI recognition terminal deployed locally on the monitoring terminal receives the insect image data, runs a pre-trained garden insect recognition model locally, classifies and recognizes the insect image data, outputs the recognition results of insect species, quantity and / or insect population density, and organizes the recognition results into structured data;

[0011] S4: The edge AI recognition terminal uploads the structured data to the cloud management platform via the wireless communication module;

[0012] S5: The cloud management platform receives and aggregates structured data from multiple monitoring terminals across the entire region, and performs pest statistical analysis, pest alarm judgment, and / or data visualization display.

[0013] According to the first aspect of the present invention, the method for identifying and monitoring invasive insects in gardens without supplemental lighting is described in step S2. The low-light imaging device without supplemental lighting has a built-in generative AI low-light imaging model. The generative AI low-light imaging model is used to automatically optimize image noise and improve the clarity of dynamic insect targets in low-light or no-light environments at night, and output full-color high-definition images.

[0014] According to the first aspect of the present invention, in the method for AI-based identification and monitoring of invasive insects in gardens without supplemental lighting, in step S3, the garden insect identification model is configured to be able to classify and identify at least three categories: invasive insects, native harmful insects, and beneficial garden insects.

[0015] According to the first aspect of the present invention, the AI-based identification and monitoring method for invasive garden insects without supplemental lighting, step S3 further includes an automatic new species sample collection step S31:

[0016] When the edge AI recognition terminal determines that the insect target in the insect image data cannot be matched and recognized by the garden insect recognition model, it automatically marks the insect target as an unknown species, and associates and records the current captured image, monitoring time, latitude and longitude location information and / or garden scene type, encapsulates it into a new species sample data package, stores it locally and uploads it to the cloud management platform.

[0017] According to the first aspect of the present invention, the AI-based identification and monitoring method for invasive garden insects without supplemental lighting, step S2 further includes an automatic update step S21 for the sticky tape:

[0018] During the monitoring process, the automatic tape replacement mechanism integrated into the monitoring terminal performs the automatic tape replacement. The automatic tape replacement mechanism includes a tape unwinding wheel, a tape take-up wheel, and a servo drive motor.

[0019] The automatic tape update is based on any of the following triggering logics:

[0020] (a) Timed update mode: According to the fixed update cycle preset by the system, the servo drive motor drives the tape take-up wheel to rotate, rolls up the used old tape to the take-up wheel, and releases the new tape on the unwind wheel to the shooting and monitoring area.

[0021] (b) Overload intelligent update mode: The edge AI recognition terminal detects the insect coverage density or effective adhesion area on the sticky insect tape in real time. When the insect coverage density exceeds the preset threshold or the effective adhesion area is lower than the preset threshold, the servo drive motor is automatically triggered to replace the tape.

[0022] According to the first aspect of the present invention, the AI-based identification and monitoring method for invasive garden insects without supplemental lighting, step S2 further includes a consumable intelligent early warning step S22:

[0023] The monitoring terminal monitors the remaining length of the sticky insect tape in real time. When the remaining length is lower than the preset warning threshold, it generates a consumable alarm information locally and uploads it to the cloud management platform. The cloud management platform then automatically generates and dispatches a consumable replacement and maintenance work order.

[0024] According to the first aspect of the present invention, in the AI-based identification and monitoring method for invasive garden insects without supplemental lighting, in step S4, the wireless communication module is a 4G / 5G dual-mode communication module, and the edge AI identification terminal uploads data according to priority levels.

[0025] High-priority data includes high-density alarm signals for invasive insects, warning signals of tape depletion, and images of samples of unknown new species;

[0026] Regular priority data includes time-segmented structured insect infestation statistics, equipment online operating status, battery level, and imaging operation logs.

[0027] According to the first aspect of the present invention, the AI-based identification and monitoring method for invasive garden insects without supplemental lighting, after step S5, further includes an algorithm iteration update step S51:

[0028] The cloud management platform receives the unknown species sample data packets uploaded by the edge AI recognition terminal. Back-end professionals manually review and identify the species attributes, incorporate the labeled new samples into the training dataset, retrain and optimize the garden insect recognition model, and remotely distribute the optimized model package to the edge AI recognition terminals of each monitoring terminal to complete the model update.

[0029] According to a second aspect of the present invention, a light-free AI identification and monitoring system for invasive garden insects is provided, including the above-described light-free AI identification and monitoring method for invasive garden insects, and further comprising...

[0030] The insect trapping module, including sticky tape and an automated tape refill mechanism, is used to trap insects and provide a monitoring carrier for imaging;

[0031] A low-light imaging device without supplemental lighting, equipped with a full-time color camera with a generative AI low-light imaging model, is used to collect images and / or videos of insect targets on the sticky tape around the clock under conditions without active supplemental lighting.

[0032] The edge AI recognition terminal module includes an edge image classification and recognition terminal with GPU computing power and a monitoring equipment housing. The edge image classification and recognition terminal is set in the monitoring equipment housing and is used to locally perform image acquisition, insect recognition, data structuring processing and local storage.

[0033] The power supply and communication module includes a solar panel, a battery pack, and a 4G / 5G dual-mode communication unit, which provides self-powered power supply and wireless data transmission.

[0034] The cloud-based management platform is used to receive and aggregate structured data uploaded by monitoring terminals across the entire region, and to perform pest statistical analysis, pest alarm assessment, manual identification of new species, and iterative management of AI models.

[0035] According to the second aspect of the present invention, the AI-based identification and monitoring system for invasive garden insects without supplemental lighting further includes a multi-scene protective installation structure module. The multi-scene protective installation structure module includes an environmentally friendly rodent-repellent, insect-proof, and corrosion-resistant coating applied to the bottom outer wall and bottom of the monitoring equipment housing. The internal electrical wires of the monitoring equipment housing are run through galvanized metal corrugated pipes or thickened nylon flame-retardant snakeskin pipes. Slow-release safety repellent materials are provided at the internal corners of the monitoring equipment housing. The seams and cable insertion holes of the monitoring equipment housing are fully sealed with sealant and rubber protective coils. The bottom of the monitoring equipment housing has a drainage hole with an insect-proof mesh.

[0036] According to the second aspect of the present invention, the AI-based identification and monitoring system for invasive insects in gardens without supplemental lighting includes a wall-mounted structure suitable for high points of trees, a low pole installation structure suitable for shrubs, and a ground-mounted installation structure suitable for lawns.

[0037] According to the second aspect of the present invention, the AI-based identification and monitoring system for invasive garden insects without supplemental lighting is provided, wherein the automated tape renewal mechanism is located within the bottom frame of the tape recycling wheel so as not to obstruct the monitoring field of view of the imaging module; the automated tape renewal mechanism is also provided with a tape consumable balance sensor for real-time monitoring of the remaining tape length and uploading it to the edge AI recognition terminal.

[0038] One of the above-described technical solutions of the present invention has at least one of the following advantages or beneficial effects:

[0039] This invention overcomes the problem of insect phototaxis interference caused by active supplemental lighting in existing technologies, and realizes natural insect monitoring without supplemental lighting or light pollution, ensuring that the monitoring data is true and consistent with the actual situation of garden ecology, and improving the accuracy of insect population density forecasting;

[0040] Real-time local insect identification and data structuring are achieved based on edge computing, supporting automatic collection and retention of new invasive insect samples. Combined with cloud-based manual review and algorithm iteration, the accuracy of invasive insect identification is continuously improved.

[0041] Through a systematic equipment installation structure and protective design, problems such as rodent and insect infestation, line damage, and moisture failure of outdoor low-profile equipment are solved, enabling long-term stable unattended monitoring of the equipment;

[0042] Construct a local garden insect dataset that distinguishes between invasive insects, native pests, and beneficial insects, while taking into account diverse application needs such as pest control, ecological surveys, scientific research and training, and forest disease prevention.

[0043] The solution adopts a solar self-powered and wireless communication integrated solution, which eliminates the need to dig up green space for wiring and connect to the municipal power grid, simplifying the on-site construction process, preventing landscape damage, and significantly reducing the cost and threshold for deployment across the entire area. Attached Figure Description

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments;

[0045] Figure 1 This is a flowchart of the AI ​​recognition and monitoring method without supplemental lighting in some embodiments of the present invention; Detailed Implementation

[0046] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0047] In the description of this invention, it should be understood that the orientation descriptions, such as up, down, front, back, left, right, etc., are based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0048] In the description of this invention, "several" means one or more, "more than" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0049] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" and "second" may explicitly or implicitly include one or more features.

[0050] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, it can be a fixed connection or a movable connection, a detachable connection or a non-detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection or a connection that allows communication between the two components; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within two components, an indirect connection, or an interaction between two components.

[0051] The following disclosure provides many different implementations or examples for different ways of implementing the present invention.

[0052] Reference Figure 1 As shown, a method for AI-based identification and monitoring of invasive insects in gardens without supplemental lighting is provided, including the following steps:

[0053] S1: Deploy monitoring terminals in garden monitoring scenarios. The monitoring terminals are equipped with sticky insect tape as insect trapping carriers and are equipped with low-light imaging equipment without supplemental lighting.

[0054] S2: Using a low-light imaging device without supplemental lighting, relying solely on ambient natural light or weak urban nighttime light, the device captures and / or records dynamic videos of insect targets on sticky tape around the clock, acquiring insect image data, under conditions where no active light source is available.

[0055] S3: The edge AI recognition terminal deployed locally on the monitoring terminal receives insect image data, runs a pre-trained garden insect recognition model locally, classifies and recognizes the insect image data, outputs the recognition results of insect species, quantity and / or insect population density, and organizes the recognition results into structured data.

[0056] S4: The edge AI recognition terminal uploads the structured data to the cloud management platform 2 via the wireless communication module;

[0057] S5: The cloud management platform 2 receives and aggregates structured data from multiple monitoring terminals across the entire domain, and performs pest statistical analysis, pest alarm judgment, and / or data visualization display.

[0058] In some embodiments of the present invention, in step S2, the low-light imaging device without supplemental lighting has a built-in generative AI low-light imaging model. The generative AI low-light imaging model is used to automatically optimize image noise and improve the clarity of dynamic insect targets in low-light or no-light environments at night, and output full-color high-definition images.

[0059] In some embodiments of the present invention, in step S3, the garden insect identification model is configured to be able to classify and identify at least three categories: invasive insects, native harmful insects, and beneficial garden insects.

[0060] In some embodiments of the present invention, step S3 further includes an automatic collection step S31 for new species samples:

[0061] When the edge AI recognition terminal determines that the insect target in the insect image data cannot be matched and recognized by the garden insect recognition model, it automatically marks the insect target as an unknown species, and records the current captured image, monitoring time, latitude and longitude location information and / or garden scene type, encapsulates it into a new species sample data package, stores it locally and uploads it to the cloud management platform2.

[0062] In some embodiments of the present invention, step S2 further includes an automatic update step S21 for the sticky tape:

[0063] During the monitoring process, the automatic tape replacement mechanism 3 integrated into the monitoring terminal performs automatic tape replacement. The automatic tape replacement mechanism 3 includes a tape unwinding wheel, a tape take-up wheel 31, and a servo drive motor.

[0064] Automatic tape updates are based on any of the following triggering logics:

[0065] (a) Timed update mode: According to the fixed update cycle preset by the system, the servo drive motor drives the tape take-up wheel 31 to rotate, and the used old tape is rolled up to the take-up wheel, while the new tape on the unwinding wheel is released to the shooting and monitoring area.

[0066] (b) Overload intelligent update mode: The edge AI recognition terminal detects the insect coverage density or effective adhesion area on the sticky insect tape in real time. When the insect coverage density exceeds the preset threshold or the effective adhesion area is lower than the preset threshold, the servo drive motor is automatically triggered to replace the tape.

[0067] In some embodiments of the present invention, step S2 further includes a consumables intelligent early warning step S22:

[0068] The monitoring terminal monitors the remaining length of the sticky insect tape in real time. When the remaining length is lower than the preset warning threshold, a consumable alarm is generated locally and uploaded to the cloud management platform 2. The cloud management platform 2 then automatically generates and dispatches a consumable replacement maintenance work order 4.

[0069] In some embodiments of the present invention, in step S4, the wireless communication module is a 4G / 5G dual-mode communication module, and the edge AI recognition terminal uploads data according to priority levels:

[0070] High-priority data includes high-density alarm signals for invasive insects, warning signals of tape depletion, and images of samples of unknown new species;

[0071] Regular priority data includes time-segmented structured insect infestation statistics, equipment online operating status, battery level, and imaging operation logs.

[0072] In some embodiments of the present invention, after step S5, an algorithm iterative update step S51 is further included:

[0073] The cloud management platform 2 receives the data package of unknown species samples uploaded by the edge AI recognition terminal. The background professionals manually review and identify the species attributes, add the labeled new samples to the training dataset, retrain and optimize the garden insect recognition model, and remotely send the optimized model package to the edge AI recognition terminals of each monitoring terminal to complete the model update.

[0074] According to a second aspect of the present invention, a light-free AI identification and monitoring system for invasive garden insects is provided, including the above-described light-free AI identification and monitoring method for invasive garden insects, and further comprising...

[0075] The insect trapping module, including sticky tape and an automated tape refill mechanism 3, is used to trap insects and provide a monitoring carrier for imaging;

[0076] The low-light imaging device 5 without supplemental lighting is equipped with a full-time color camera with a generative AI low-light imaging model, which is used to collect images and / or videos of insect targets on sticky tape in all weather conditions without active supplemental lighting.

[0077] The edge AI recognition terminal module includes an edge image classification and recognition terminal with GPU computing power and a monitoring equipment housing. The edge image classification and recognition terminal is set inside the monitoring equipment housing and is used to perform image acquisition, insect recognition, data structuring processing and local storage locally.

[0078] The power supply and communication module includes a solar panel, a battery pack, and a 4G / 5G dual-mode communication unit, which provides self-powered power supply and wireless data transmission.

[0079] The cloud management platform 2 is used to receive and aggregate structured data uploaded by monitoring terminals across the entire domain, and to perform pest statistical analysis, pest alarm judgment, manual identification of new species, and AI model iteration management.

[0080] In some embodiments of the present invention, a multi-scenario protective installation structure module is also included. The multi-scenario protective installation structure module includes an environmentally friendly rodent-repellent, insect-proof, and corrosion-resistant coating on the bottom outer wall and bottom of the monitoring equipment housing; internal wires of the monitoring equipment housing are run through galvanized metal corrugated pipes or thickened nylon flame-retardant braided tubing; slow-release safety repellent materials are provided at the internal corners of the monitoring equipment housing; the seams of the monitoring equipment housing and the cable passage holes are fully sealed with sealant and rubber protective coils; and a drainage hole with an insect-proof net is provided at the bottom of the monitoring equipment housing.

[0081] In some embodiments of the present invention, the installation forms of the monitoring terminal include wall-mounted structures suitable for high points of trees, low pole installation structures suitable for shrubs, and underground installation structures suitable for lawns.

[0082] In some embodiments of the present invention, the automated tape renewal mechanism 3 is set inside the bottom frame of the tape recycling wheel 31 so as not to obstruct the monitoring field of the imaging module; the automated tape renewal mechanism 3 is also equipped with a tape consumable balance sensor 32 for real-time monitoring of the remaining tape length and uploading it to the edge AI recognition terminal.

[0083] Example 1

[0084] This invention provides an AI-based identification and monitoring system for invasive insects in gardens without supplemental lighting. The system consists of a three-layer structure: a front-end distributed monitoring layer, an edge computing layer, and a cloud management platform.

[0085] The front-end distributed monitoring layer includes at least three types of monitoring terminals, depending on the different garden scenarios: wall-mounted monitoring terminals suitable for high points of trees, low pole monitoring terminals suitable for shrubs, and ground-buried monitoring terminals suitable for lawns. These are deployed throughout the entire area to form a grid-based monitoring network.

[0086] Each monitoring terminal integrates an insect trapping module and a low-light imaging device without supplemental illumination. The insect trapping module includes sticky insect tape and an automated tape replacement mechanism. The sticky insect tape is yellow and flexible, suspended at the monitoring point to trap winged, piercing-sucking invasive insects. The automated tape replacement mechanism 3 is located within the bottom frame of the tape recovery wheel 31 to avoid obstructing the monitoring field of view of the imaging module. It includes a tape unwinding wheel, a tape recovery wheel 31, and a servo drive motor for automatic replacement of the sticky insect tape.

[0087] The automated tape renewal mechanism 3 has two triggering logics: one is the timed renewal mode, where the system can preset a fixed renewal cycle (e.g., every 24 hours or 48 hours). The servo motor drives the tape recycling wheel 31 to rotate at a constant speed, rolling up the old tape covered with insects and dust to the recycling wheel, and simultaneously releasing a brand new clean tape to the shooting and monitoring area; the other is the overload intelligent renewal mode, where the edge AI recognition terminal identifies the insect coverage density on the tape in real time. When it detects that too many insects have accumulated or the effective adhesion area of ​​the tape is insufficient (e.g., the effective adhesion area is less than 70% of the total tape area), the servo motor is automatically triggered to rotate, and the tape is dynamically renewed in an instant.

[0088] The low-light imaging device 5 without supplemental lighting includes a full-time color camera with a built-in generative AI low-light imaging model. This camera does not have any active light source module, relying entirely on ambient natural light and the low light of urban nighttime environments for imaging. The generative AI low-light imaging model can automatically optimize image noise and improve the clarity of dynamic insect targets in low-light or no-light environments, outputting full-color high-definition images. The camera captures images of insects on sticky tape at preset intervals (e.g., every 30 minutes) and can also record dynamic video.

[0089] The edge AI recognition terminal module includes a monitoring equipment enclosure and an edge image classification and recognition terminal housed within the enclosure. The edge image classification and recognition terminal is equipped with GPU computing power for local image acquisition, insect identification, data structuring processing, and local storage. The enclosure integrates a power supply and communication module, including a solar panel (which can be installed on the top or side of the enclosure), a battery pack, and a 4G / 5G dual-mode communication unit for self-powering and wireless data transmission.

[0090] To cope with complex outdoor environments, the monitoring terminal is also equipped with a multi-scenario protective installation structure module. This module specifically includes: the lower half of the outer wall and bottom of the monitoring equipment enclosure are coated with environmentally friendly rodent-repellent, insect-proof, and corrosion-resistant paint; all internal wiring is run through galvanized metal corrugated conduit or thickened nylon flame-retardant braided tubing; slow-release safety repellent materials are suspended or placed in the corners of the enclosure; all joints and cable holes are fully sealed with sealant and rubber protective coils; and a drainage hole with an insect-proof mesh is located at the bottom of the enclosure. This combined design effectively prevents rodents from gnawing on the wiring, small insects from entering, and moisture from entering the enclosure.

[0091] Cloud Management Platform 2 is deployed on a cloud server to receive and aggregate structured data uploaded by monitoring terminals across the entire domain, and to perform pest statistical analysis, pest alarm assessment, manual identification of new species, and AI model iteration management.

[0092] Example 2

[0093] Please refer to Figure 1 This invention provides an AI-based identification and monitoring method for invasive insects in gardens without supplemental lighting. This method is based on the system described in Embodiment 1 and mainly includes the following steps:

[0094] S1: Equipment deployment and initialization.

[0095] Select a suitable installation structure to deploy the monitoring terminal based on the monitoring scenario (tree high points, shrubbery, or lawn). Install insect-adhesive tape on the automatic rewinding mechanism within the device's field of view; check the alarm and fault information in the historical logs; complete the debugging and calibration of all parameters, including solar power supply, 4G / 5G communication, AI recognition parameters, tape replacement cycle, insect overload threshold, and consumable remaining warning threshold.

[0096] S2: All-weather imaging without supplemental lighting and automatic updating of insect-sticking tape.

[0097] In terms of imaging acquisition, the all-weather color camera equipped with a generative AI low-light imaging model operates continuously 24 hours a day, relying solely on the low light of the natural environment. It captures images of the adhesive surface at preset intervals (such as every 30 minutes) and can also record dynamic video. There is no artificial light source throughout the process, so it will not attract photophilic insects, ensuring that the activity of insect groups closely matches the natural situation.

[0098] Regarding automatic tape replacement, the automated tape replacement mechanism operates in two modes: ① Timed mode: When the system reaches the preset cycle (e.g., every 24 hours), the servo motor drives the tape recycling wheel to rotate at a constant speed, collecting the old tape and releasing the new tape; ② Overload intelligent mode: The edge AI terminal calculates the insect coverage ratio in real time. When too many insects accumulate and the effective adhesive area is lower than the set threshold, the servo motor is immediately activated to replace the tape.

[0099] Meanwhile, the remaining length of the tape is monitored in real time by the tape consumable balance sensor 32. When the balance is lower than the warning value, an early warning signal is generated locally and uploaded to the cloud management platform 2. The cloud management platform 2 automatically generates a consumable replacement maintenance work order 4 and pushes it to the maintenance personnel in the corresponding area.

[0100] S3: Local AI parsing, sample labeling, and data structuring processing on edge GPU terminals.

[0101] The edge AI recognition terminal receives image data captured by a camera and runs a pre-trained garden insect recognition model locally. This model can classify and identify at least three categories of insects: invasive insects (such as the fall webworm and the sycamore lace bug), native harmful insects, and beneficial garden insects (such as natural enemies). The terminal accurately counts the number of each type of insect and calculates the insect population density at each location.

[0102] For unfamiliar insects that the model cannot match and identify, the new species sample collection step S31 is automatically executed: the insect is marked as an unknown species, and associated with the current captured image, GPS location coordinates, precise time, and greening scene type, packaged into a new species sample data package, encrypted and stored locally and awaiting upload.

[0103] All images, raw videos, recognition results, and sample files are encrypted and stored locally to prevent data loss in the event of network interruption.

[0104] S4: Hierarchical wireless transmission to the cloud management platform 2.

[0105] Edge terminals upload data according to priority levels via a 4G / 5G dual-mode communication module. High-priority data includes: high-density alarm signals for invasive insects, warning signals for tape depletion, and images of samples of unknown new species; regular-priority data includes: time-segmented structured insect infestation statistics, equipment online status, battery level, and imaging operation logs. When the network is weak, the terminal uploads data in batches with a delay, and automatically re-uploads the cached data after the network is restored.

[0106] S5: Cloud management platform analysis, manual review and algorithm iteration.

[0107] Cloud management platform 2 receives and aggregates structured data from multiple monitoring terminals across the entire domain, and performs the following operations:

[0108] (1) Overall insect situation assessment: Generate heat map of invasive insect distribution and population time change curve; when the insect population density exceeds the standard, automatically push prevention and control alarm to the garden management backend and maintenance mobile terminal.

[0109] (2) Manual identification of new species: The plant protection technicians in the background retrieve the images of unknown samples, manually determine the species attributes (such as quarantine invasive species, local rare species or common beneficial insects), and complete the accurate labeling of the samples.

[0110] (3) AI model iteration optimization (step S51): The newly added insect samples with manual annotations are added to the training dataset, and the recognition model is retrained and optimized; the optimized lightweight model package is remotely distributed to all edge GPU terminals to complete the synchronous upgrade of the algorithm of the whole domain devices.

[0111] S6: Long-term data accumulation and multi-scenario big data application.

[0112] (1) The system continuously outputs standardized insect population density monitoring reports and pest classification early warning notifications to guide precise targeted pest control and ecological prevention and control;

[0113] (2) Accumulate a database of local insect phenological cycles, population distribution, and invasive species dispersal trajectories year by year;

[0114] (3) Open data interfaces to support diverse application scenarios such as garden ecological surveys, forest tree quarantine and epidemic prevention, agricultural and forestry scientific research and college teaching and training.

[0115] The equipment executes steps S2 to S6 cyclically 24 / 7 throughout the year, enabling unattended long-term fixed-site monitoring.

[0116] The above is a further detailed description of the present invention and should not be considered as a limitation on the specific implementation of the present invention. For those skilled in the art, simple deductions or substitutions without departing from the concept of the present invention are all within the protection scope of the present invention.

Claims

1. A method for AI-based identification and monitoring of invasive insects in gardens without supplemental lighting, characterized in that, Includes the following steps: S1: In the garden monitoring scenario, a monitoring terminal is deployed. The monitoring terminal is equipped with sticky insect tape as an insect trapping carrier and is equipped with a low-light imaging device without supplemental lighting. S2: Using the aforementioned low-light imaging device without supplementary lighting, under conditions without active light source supplementary lighting, relying only on ambient natural light or weak urban night light, the insect targets on the sticky tape are captured in real time and / or recorded in dynamic video to obtain insect image data. S3: The edge AI recognition terminal deployed locally on the monitoring terminal receives the insect image data, runs a pre-trained garden insect recognition model locally, classifies and recognizes the insect image data, outputs the recognition results of insect species, quantity and / or insect population density, and organizes the recognition results into structured data; S4: The edge AI recognition terminal uploads the structured data to the cloud management platform (2) through the wireless communication module. S5: The cloud management platform (2) receives and aggregates structured data from multiple monitoring terminals across the entire domain, and performs pest statistical analysis, pest alarm judgment and / or data visualization display.

2. The AI-based identification and monitoring method for invasive garden insects without supplemental lighting as described in claim 1, characterized in that, In step S2, the low-light imaging device without supplemental lighting has a built-in generative AI low-light imaging model. The generative AI low-light imaging model is used to automatically optimize image noise and improve the clarity of dynamic insect targets in low-light or nighttime light-free environments, and output full-color high-definition images.

3. The AI-based identification and monitoring method for invasive garden insects without supplemental lighting as described in claim 1, characterized in that, In step S3, the garden insect identification model is configured to be able to classify and identify at least three categories: invasive insects, native harmful insects, and beneficial garden insects.

4. The AI-based identification and monitoring method for invasive garden insects without supplemental lighting according to claim 1, characterized in that, Step S3 also includes an automatic new species sample collection step S31: When the edge AI recognition terminal determines that the insect target in the insect image data cannot be matched and recognized by the garden insect recognition model, it automatically marks the insect target as an unknown species, and associates and records the current captured image, monitoring time, latitude and longitude location information and / or garden scene type, encapsulates it into a new species sample data package, stores it locally and uploads it to the cloud management platform (2).

5. The AI-based identification and monitoring method for invasive garden insects without supplemental lighting according to claim 1, characterized in that, Step S2 also includes an automatic update step S21 for the sticky tape: During the monitoring process, the automatic tape replacement is performed by the automated tape replacement mechanism (3) integrated into the monitoring terminal. The automated tape replacement mechanism (3) includes a tape unwinding wheel, a tape take-up wheel (31), and a servo drive motor. The automatic tape update is based on any of the following triggering logics: (a) Timed update mode: According to the fixed update cycle preset by the system, the servo drive motor drives the tape recycling wheel (31) to rotate, roll up the used old tape to the recycling wheel, and release the new tape on the unwinding wheel to the shooting monitoring area. (b) Overload intelligent update mode: The edge AI recognition terminal detects the insect coverage density or effective adhesion area on the sticky insect tape in real time. When the insect coverage density exceeds the preset threshold or the effective adhesion area is lower than the preset threshold, the servo drive motor is automatically triggered to replace the tape.

6. The AI-based identification and monitoring method for invasive garden insects without supplemental lighting according to claim 5, characterized in that, Step S2 also includes a consumables intelligent early warning step S22: The monitoring terminal monitors the remaining length of the sticky insect tape in real time. When the remaining length is lower than the preset warning threshold, it generates a consumable alarm information locally and uploads it to the cloud management platform (2). The cloud management platform (2) automatically generates and dispatches a consumable replacement maintenance work order (4).

7. The AI-based identification and monitoring method for invasive garden insects without supplemental lighting according to claim 1, characterized in that, In step S4, the wireless communication module is a 4G / 5G dual-mode communication module, and the edge AI recognition terminal uploads data according to priority levels. High-priority data includes high-density alarm signals for invasive insects, warning signals of tape depletion, and images of samples of unknown new species; Regular priority data includes time-segmented structured insect infestation statistics, equipment online operating status, battery level, and imaging operation logs.

8. The AI-based identification and monitoring method for invasive garden insects without supplemental lighting according to claim 1, characterized in that, Step S5 also includes an algorithm iterative update step S51: The cloud management platform (2) receives the unknown species sample data packet uploaded by the edge AI recognition terminal. The background professionals manually review and identify the species attributes, include the labeled new sample into the training dataset, retrain and optimize the garden insect recognition model, and remotely send the optimized model package to the edge AI recognition terminal of each monitoring terminal to complete the model update.

9. A non-supplementary AI identification and monitoring system for invasive insects in gardens, characterized in that, The method for AI-based identification and monitoring of invasive garden insects without supplemental lighting, as described in any one of claims 1-8, further includes: The insect trapping module, including sticky tape and an automated tape refill mechanism (3), is used to trap insects and provide a monitoring carrier for imaging; The low-light imaging device without supplemental lighting (5) is equipped with a full-time color camera with a generative AI low-light imaging model, which is used to collect images and / or videos of insect targets on the sticky tape all day long under conditions without active supplemental lighting. The edge AI recognition terminal module includes an edge image classification and recognition terminal with GPU computing power and a monitoring equipment housing. The edge image classification and recognition terminal is set in the monitoring equipment housing and is used to locally perform image acquisition, insect recognition, data structuring processing and local storage. The power supply and communication module includes a solar panel, a battery pack, and a 4G / 5G dual-mode communication unit, which provides self-powered power supply and wireless data transmission. The cloud management platform (2) is used to receive and summarize the structured data uploaded by the full-domain monitoring terminals, and to perform pest statistical analysis, pest alarm judgment, new species manual identification and AI model iteration management.

10. The AI-based identification and monitoring system for invasive garden insects without supplemental lighting according to claim 9, characterized in that, It also includes a multi-scenario protective installation structure module, wherein the bottom outer wall and bottom of the monitoring equipment box are coated with environmentally friendly rodent-repellent, insect-proof, and corrosion-resistant coatings; the internal wires of the monitoring equipment box are run through galvanized metal corrugated pipes or thickened nylon flame-retardant braided tubing; slow-release safety repellent materials are provided at the internal corners of the monitoring equipment box; the joints and cable holes of the monitoring equipment box are fully sealed with sealant and rubber protective coils; and the bottom of the monitoring equipment box has a drainage hole with an insect-proof net.

11. The AI-based identification and monitoring system for invasive garden insects without supplemental lighting according to claim 9, characterized in that, The monitoring terminals can be installed in various ways, including wall-mounted structures suitable for treetops, low-pole installation structures suitable for shrubbery, and underground installation structures suitable for lawns.

12. The AI-based identification and monitoring system for invasive garden insects without supplemental lighting according to claim 9, characterized in that, The automated tape renewal mechanism (3) is set in the bottom frame of the tape recycling wheel (31) so as not to obstruct the monitoring field of the imaging module; the automated tape renewal mechanism (3) is also equipped with a tape consumable balance sensor (32) for real-time monitoring of the remaining tape length and uploading it to the edge AI recognition terminal.

Images