Plant quarantine pest intelligent trapping device integrated with image recognition function

Abstract

The present application belongs to the technical field of plant quarantine, and particularly relates to a plant quarantine pest intelligent trapping device integrated with image recognition functions, which comprises a trapping main body module, an image acquisition and recognition module, an intelligent control module, a data transmission module and a power supply module, and each module is electrically connected with the intelligent control module through a line. The present application realizes integrated intelligent operation: the functions of trapping, image acquisition, real-time recognition, statistics, early warning and data uploading are integrated, the whole-process monitoring of quarantine pests can be completed without manual intervention, the cost of manual inspection and recognition is greatly reduced, and the work efficiency of quarantine is improved.

Description

Technical Field

[0001] This invention belongs to, but is not limited to, the field of plant quarantine technology, and particularly relates to an intelligent trapping device for plant quarantine pests with integrated image recognition function. Background Technology

[0002] Plant quarantine pests pose a significant threat to agricultural production and ecological security, primarily introduced through international trade, transportation, and passenger transport. The "List of Quarantine Pests Entering the People's Republic of China" lists 178 species (genera) of quarantine pests, the "National List of Quarantine Pests of Agricultural Plants" lists 9 species, and the "National List of Quarantine Pests of Forestry Plants" lists 10 species. Many of these quarantine pests are highly destructive invasive alien species, making early, efficient, continuous, and automated monitoring methods fundamental for precise control. Traditional pest monitoring relies heavily on regular manual inspections or manual counting of pests in traps. This method is labor-intensive, inefficient, and lacks timeliness, failing to meet the requirements of large-scale, long-term, and continuous monitoring in modern agriculture. Therefore, domestic and international research and industry have long been committed to integrating automated sensing technology, image recognition technology, and remote communication technology into pest trapping and monitoring systems to improve the automation level and response speed of monitoring.

[0003] In recent years, with the rapid development of artificial intelligence, big data, and the Internet of Things (IoT) technologies, some image-sensing-based trapping and monitoring systems have begun to enter the research and application stage. A literature review indicates that integrating image sensors into traps, transmitting captured image data via wireless links, and applying machine learning algorithms at central stations or edge computing nodes to identify and count pests from the images is a trend that improves monitoring efficiency and reduces labor costs. Examples of this research include using low-power wireless image sensing nodes to periodically capture images of pests inside traps and extracting individual population information through image processing algorithms to achieve automatic statistical and temporal analysis of pest populations.

[0004] In industrial practice, some intelligent trapping and monitoring devices have already implemented image acquisition and remote transmission functions. For example, some solar-powered intelligent trapping devices can automatically photograph pests in the trapping board or trapping chamber and upload the images to a remote monitoring platform. The platform, combined with simple image analysis algorithms, can count the number of pests and distinguish their basic species. These devices generally integrate a trapping mechanism, a camera module, a communication unit, and a power supply unit, which can reduce the frequency of manual on-site inspections to a certain extent and support centralized management of trapping data through the Internet of Things.

[0005] Despite the advancements in technology towards intelligentization, significant shortcomings remain. First, existing systems primarily focus on image acquisition and uploading, lacking the real-time identification capabilities for pest species and quantities at the field site. Image data typically needs to be transmitted back to a remote server for processing, leading to response delays and hindering automatic adjustment and early warning control at the device level. This limitation restricts the system's ability to respond to rapid changes in pest populations. Second, existing devices often employ single light sources or passive insect-attracting methods, resulting in limited effectiveness against different pest species and difficulty in dynamically adjusting trapping parameters based on target species. Intelligent adjustment capabilities in areas such as pheromone release management and trapping spectrum selection are also weak. Third, existing products typically rely on a single wireless communication mode, lacking multi-mode communication switching capabilities and being susceptible to environmental and network conditions. Furthermore, more comprehensive power supply and management strategies are needed for overall energy management and ensuring continuous operation of the equipment, especially in monitoring areas far from power supplies.

[0006] Therefore, while existing technologies have made some progress in achieving "automated image acquisition and remote transmission," significant technical bottlenecks remain in areas such as "on-site image recognition and decision control," "multi-mode trapping parameter adjustment," "multi-communication mode integration," and "energy collaborative supply guarantee." These limitations prevent the formation of a comprehensive solution integrating effective local identification, intelligent control, and stable remote management. These shortcomings directly affect monitoring accuracy, timeliness, and the equipment's on-site adaptability. Summary of the Invention

[0007] To address the problems existing in the prior art, this invention provides an intelligent trapping device for plant quarantine pests that integrates image recognition function.

[0008] This invention is implemented as follows: an intelligent trapping device for plant quarantine pests with integrated image recognition function, the device comprising:

[0009] The system includes a trapping module, an image acquisition and recognition module, an intelligent control module, a data transmission module, and a power supply module. Each module is electrically connected to the intelligent control module via a line.

[0010] The main trapping module includes a trapping cavity, a multi-band trapping light source, a specific pheromone release component, and a pest collection trough. The trapping cavity is a cylindrical structure with openings at the top and bottom, and the inner wall is provided with an anti-slip coating. The multi-band trapping light source is embedded in the upper edge of the trapping cavity and can switch between ultraviolet light, visible light, and infrared light bands. The specific pheromone release component is installed in the middle of the trapping cavity, which stores pheromones through a sealed chamber and releases them quantitatively. The pest collection trough is detachably connected to the bottom of the trapping cavity.

[0011] The image acquisition and recognition module includes a high-definition industrial camera, a supplementary lighting component, and an image recognition unit. The lens of the high-definition industrial camera is oriented towards the pest collection trough. The supplementary lighting component starts and stops synchronously with the camera to provide uniform, shadow-free illumination. The image recognition unit has a built-in trained convolutional neural network model, which can preprocess, extract features, and identify the species and quantity of pests in the images acquired by the camera, with a recognition accuracy of no less than 95%.

[0012] The intelligent control module uses a microcontroller as the core controller. It can receive the recognition results from the image recognition unit, automatically adjust the band, brightness and pheromone release rate of the multi-band trapping light source, trigger a local sound and light warning when a quarantine pest is detected, and send a data upload command to the data transmission module.

[0013] The data transmission module supports 4G / 5G, WiFi and LoRa multi-mode communication, and can upload pest species, quantity, identification time, device location and operating status data to the cloud management platform in real time. It can also receive parameter adjustment instructions from the cloud.

[0014] The power supply module includes a solar panel, a battery, and a charging management unit. The solar panel is installed at an angle on the top of the trapping main module. The battery is a rechargeable lithium battery. The charging management unit enables complementary power supply from solar energy and mains power to ensure continuous operation of the equipment.

[0015] Furthermore, the wavelength adjustment range of the multi-band trapping light source is 365nm - 850nm, and it supports preset band combinations according to the target quarantine pest species. Automatic switching is achieved through the intelligent control module, and the brightness of the light source can be steplessly adjusted within the range of 0 - 1000 lux.

[0016] Furthermore, the specific pheromone release component includes multiple independent sealed chambers, which can simultaneously store 2-4 different types of quarantine pest pheromones. The release rate can be adjusted from 0.1 to 1.0 ml / h, and the release amount can be dynamically adjusted by the intelligent control module based on ambient temperature and humidity parameters.

[0017] Furthermore, the convolutional neural network model of the image recognition unit adopts the ResNet-50 architecture and has been trained and optimized with at least 50,000 positive and negative sample images of different quarantine pests. It can identify at least 20 common quarantine pests with a recognition response time of no more than 2 seconds.

[0018] Furthermore, the intelligent control module also has a built-in GPS positioning unit, which can collect device location information in real time and upload it to the cloud management platform. It supports multi-device network management and automatically sends fault warning information when the device malfunctions (power failure, communication interruption, component damage).

[0019] Furthermore, the pest collection trough is made of transparent and wear-resistant material, and the trough wall is equipped with scale lines to facilitate manual verification of the number of pests. The connection between the collection trough and the trapping chamber is a snap-on type, and the disassembly time does not exceed 30 seconds.

[0020] Furthermore, the supplementary lighting component adopts a ring-shaped LED supplementary light, which is set around the lens of a high-definition industrial camera. The color temperature adjustment range is 3000K - 6500K, and the supplementary lighting angle can be adjusted through a mechanical structure to ensure that the collected images of pests are free of reflection and distortion.

[0021] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0022] 1. Achieve integrated intelligent operation: Integrates functions such as trapping, image acquisition, real-time identification, statistics, early warning, and data uploading, and can complete the entire process of monitoring quarantine pests without human intervention, greatly reducing the cost of manual inspection and identification and improving the efficiency of quarantine work.

[0023] 2. High accuracy and rapid response: Utilizing an optimized convolutional neural network model trained on a large scale, the recognition accuracy is no less than 95%, and the response time is no more than 2 seconds. It can quickly and accurately identify a variety of quarantine pests, avoiding the errors and delays of manual identification.

[0024] 3. Highly targeted and adaptive trapping: The combination of multi-band trapping light sources and specific pheromones can target and trap pests. At the same time, the trapping parameters are dynamically adjusted based on the recognition results and environmental parameters to improve trapping efficiency and adapt to different pest species and environmental scenarios.

[0025] 4. Stable operation and wide adaptability: It adopts a complementary power supply of solar energy and mains power, with strong endurance and can adapt to complex outdoor environments; it supports multi-mode communication and multi-device networking, and can be widely used in plant quarantine work in different scenarios such as farmland, ports, nurseries, and orchards. Attached Figure Description

[0026] Figure 1 This is a structural diagram of an intelligent trapping device for plant quarantine pests with integrated image recognition function provided in an embodiment of the present invention;

[0027] Figure 2 This is a structural diagram of the trapping main module provided in an embodiment of the present invention;

[0028] Figure 3 This is a structural diagram of the image acquisition and recognition module provided in an embodiment of the present invention;

[0029] Figure 4 This is a structural diagram of the intelligent control module provided in an embodiment of the present invention;

[0030] In the diagram: 1. Trapping main module; 2. Image acquisition and recognition module; 3. Intelligent control module; 4. Data transmission module; 5. Power supply module; 6. Trapping cavity; 7. Multi-band trapping light source; 8. Specific pheromone release component; 9. Pest collection trough; 10. High-definition industrial camera; 11. Supplemental lighting component; 12. Image recognition unit; 13. Microcontroller. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0032] like Figure 1 As shown, this embodiment of the invention provides an intelligent trapping device for plant quarantine pests with integrated image recognition function. The device includes:

[0033] The system consists of a trapping module 1, an image acquisition and recognition module 2, an intelligent control module 3, a data transmission module 4, and a power supply module 5. Each module is electrically connected to the intelligent control module 3 via a line.

[0034] like Figure 2 As shown, the main trapping module 1 includes a trapping cavity 6, a multi-band trapping light source 7, a specific pheromone release component 8, and a pest collection trough 9. The trapping cavity 6 is a cylindrical structure with openings at the top and bottom, and the inner wall is provided with an anti-slip coating. The multi-band trapping light source 7 is embedded in the upper edge of the trapping cavity 6 and can switch between ultraviolet light, visible light, and infrared light bands. The specific pheromone release component 8 is installed in the middle of the trapping cavity 6, which stores pheromones through a sealed chamber and releases them quantitatively. The pest collection trough 9 is detachably connected to the bottom of the trapping cavity 6.

[0035] like Figure 3 As shown, the image acquisition and recognition module 2 includes a high-definition industrial camera 10, a supplementary lighting component 11, and an image recognition unit 12. The lens of the high-definition industrial camera 10 is directed toward the pest collection trough 9. The supplementary lighting component 11 starts and stops synchronously with the camera to provide uniform, shadow-free illumination. The image recognition unit 12 has a built-in convolutional neural network model that has been trained. It can preprocess, extract features, and identify the types and quantities of pests in the images acquired by the camera, with an accuracy of not less than 95%.

[0036] The intelligent trapping device for plant quarantine pests with integrated image recognition function provided in this embodiment of the invention forms a closed-loop working mechanism with "trapping behavior guidance - standardized imaging - intelligent recognition - adaptive control" as its core in terms of overall structural design and functional coordination. It is not a simple superposition of trapping device and image recognition device, but achieves highly reliable, quantifiable and traceable monitoring of quarantine pests through deep coupling of spatial structure, temporal control and algorithm logic.

[0037] In actual operation, the intelligent control module 3, as the core of the system, controls the trapping module 1 to enter the working state according to the preset monitoring strategy. Under the scheduling of the intelligent control module 3, the multi-band trapping light source 7 switches between ultraviolet, visible, and infrared light bands according to a preset sequence, forming a photochemical synergistic attraction mechanism with the specific pheromone release component 8. Specifically, the multi-band trapping light source 7 is responsible for achieving broad-spectrum attraction of pests with different phototactic behaviors, while the specific pheromone release component 8 achieves targeted screening of the trapped objects by quantitatively releasing pheromones highly matched with the target quarantine pests, thereby reducing the probability of non-target insects entering the trapping chamber 6 from the source.

[0038] The trapping chamber 6 adopts a cylindrical structure with openings at the top and bottom, and an anti-slip coating on the inner wall, making it difficult for attracted pests to escape after entering. Under the influence of gravity, they naturally fall into the pest collection trough 9. This structure allows the pests to complete a stable spatial posture transition before entering the collection area, creating consistent imaging conditions for subsequent image acquisition. The detachable connection between the pest collection trough 9 and the trapping chamber 6 not only facilitates cleaning and maintenance but also ensures that their position remains fixed during imaging, avoiding recognition errors caused by vibration or displacement.

[0039] Once the pests enter the pest collection trough 9, the image acquisition and recognition module 2 is activated under the trigger of the intelligent control module 3. The supplementary lighting component 11 works synchronously with the high-definition industrial camera 10, eliminating the impact of ambient light changes on image quality through uniform, shadowless illumination, ensuring stable brightness, contrast, and background consistency in the acquired images. The high-definition industrial camera 10 is fixedly oriented towards the pest collection trough 9, ensuring that each pest is captured within a preset field of view and scale, thus avoiding the uncertainty of posture and scale caused by traditional open-environment shooting.

[0040] After receiving image data from the high-definition industrial camera 10, the image recognition unit 12 sequentially performs image preprocessing, feature extraction, and species and quantity identification based on a convolutional neural network model. Since the trapping structure physically constrains the spatial location and imaging conditions of the pests, the image recognition unit 12 can achieve high-confidence identification without relying on complex background segmentation or multi-target tracking algorithms, thus significantly improving recognition accuracy and system stability. The identification results are integrated by the intelligent control module 3 and then sent to a remote terminal via the data transmission module 4 for real-time monitoring and early warning.

[0041] This invention constructs an intelligent trapping mechanism with inherent functional coupling and logical closed loop through the overall synergy between trapping structure design, coordinated attraction of light source and pheromone, physical constraints of imaging conditions, and intelligent recognition algorithm. This makes it impossible for the dispersed trapping, image recognition and control methods in the prior art to achieve the technical effect of this invention through simple combination, and has significant overall inventiveness.

[0042] like Figure 4 As shown, the intelligent control module 3 uses a single-chip microcomputer 13 as the core controller, which can receive the recognition results of the image recognition unit 12, automatically adjust the band, brightness and pheromone release rate of the multi-band trapping light source, trigger a local sound and light warning when a quarantine pest is detected, and send a data upload command to the data transmission module 4.

[0043] The data transmission module 4 supports 4G / 5G, WiFi and LoRa multi-mode communication, and can upload pest species, quantity, identification time, device location and operating status data to the cloud management platform in real time. At the same time, it can receive parameter adjustment instructions issued by the cloud.

[0044] The power supply module 5 includes a solar panel, a battery, and a charging management unit. The solar panel is installed at an angle on the top of the trapping main module. The battery is a rechargeable lithium battery. The charging management unit enables complementary power supply from solar energy and mains power to ensure continuous operation of the equipment.

[0045] The wavelength adjustment range of the multi-band trapping light source 7 is 365nm - 850nm. It supports preset band combinations according to the target quarantine pest species and can automatically switch through the intelligent control module 3. The brightness of the light source can be steplessly adjusted in the range of 0 - 1000 lux.

[0046] The specific pheromone release component 8 includes multiple independent sealed chambers, which can simultaneously store 2-4 different types of quarantine pest pheromones. The release rate can be adjusted from 0.1 to 1.0 ml / h, and the release amount can be dynamically adjusted by the intelligent control module based on ambient temperature and humidity parameters.

[0047] The convolutional neural network model of the image recognition unit 12 adopts the ResNet-50 architecture and has been trained and optimized with at least 50,000 positive and negative sample images of different quarantine pests. It can identify at least 20 common quarantine pests with a recognition response time of no more than 2 seconds.

[0048] The intelligent control module 3 also has a built-in GPS positioning unit, which can collect device location information in real time and upload it to the cloud management platform. It supports multi-device network management and automatically sends fault warning information when the device malfunctions (power failure, communication interruption, component damage).

[0049] The pest collection trough 9 is made of transparent and wear-resistant material. The trough wall is marked with scale lines to facilitate manual verification of the number of pests. The connection between the collection trough and the trapping chamber is a snap-on type, and the disassembly time does not exceed 30 seconds.

[0050] The supplementary lighting component 11 uses a ring-shaped LED supplementary light, which is set around the lens of a high-definition industrial camera. The color temperature adjustment range is 3000K - 6500K, and the supplementary lighting angle can be adjusted through a mechanical structure to ensure that the collected images of pests are free from reflection and distortion.

[0051] In this embodiment of the invention, the intelligent trapping device for plant quarantine pests with integrated image recognition function constructs an adaptive trapping and recognition collaborative working mechanism with intelligent control module 3 as the core through deep coupling of multiple modules. Its overall operation is not the independent operation or linear series of each functional module, but rather the dynamic linkage of trapping parameters, recognition results and operating status is achieved through data-driven methods.

[0052] During equipment operation, the power supply module 5 prioritizes powering the battery through the solar panels tilted at the top, and the charging management unit intelligently switches between solar and mains power, thus providing a stable power foundation for long-term continuous operation in the field. Under the premise of stable power supply, the microcontroller 13 in the intelligent control module 3 enters the working state and initializes the operating parameters of the multi-band trapping light source 7, the specific pheromone release component 8, the image acquisition and recognition module 2, and the data transmission module 4.

[0053] During the trapping phase, the intelligent control module 3, based on a preset target quarantine pest strategy, controls the multi-band trapping light source 7 to select corresponding band combinations within the range of 365nm to 850nm, and continuously adjusts the brightness of the light source, enabling the optical attraction and the specific pheromone release component 8 to work synergistically. With the cooperation of multiple independent sealed chambers, the specific pheromone release component 8 quantitatively releases the corresponding pheromone into the trapping chamber 6 according to the release rate set by the microcontroller 13, while dynamically correcting for environmental temperature and humidity parameters, thereby avoiding the trapping efficiency fluctuations caused by traditional constant release methods.

[0054] The lured pests enter the trapping chamber 6 under the influence of multimodal trapping and fall into the pest collection trough 9 under the constraint of gravity and the anti-slip structure of the inner wall. The position, transparent material, and scale design of the pest collection trough 9 ensure that the pests are in a stable, visible, and uniformly sized state in physical space, providing standardized sample conditions for subsequent imaging and identification.

[0055] When pests enter the pest collection trough 9, the intelligent control module 3 triggers the image acquisition and recognition module 2 to start. The supplementary lighting component 11 provides uniform ring-shaped supplementary lighting in sync with the high-definition industrial camera 10. By adjusting the color temperature and illumination angle, it eliminates the effects of reflection and shadow, ensuring that the imaging conditions are consistent with the training samples. The image data acquired by the high-definition industrial camera 10 is sent to the image recognition unit 12 in real time. The image recognition unit 12, based on a ResNet50 architecture convolutional neural network model, completes image preprocessing, feature extraction, and pest species and quantity identification, outputting the recognition results within 2 seconds.

[0056] After the identification result is received by the intelligent control module 3, it serves as a feedback signal to adjust the band combination, brightness parameters, and release rate of the specific pheromone release component 8 of the multi-band trapping light source 7, thereby forming a closed-loop control mechanism that drives the optimization of the trapping strategy based on the identification result. When a quarantine pest is identified, the intelligent control module 3 simultaneously triggers a local audible and visual alarm and uploads the pest species, quantity, identification time, equipment location, and operating status to the cloud management platform via the data transmission module 4.

[0057] Meanwhile, the built-in GPS positioning unit of the intelligent control module 3 collects the device's location in real time, enabling the cloud platform to achieve spatial distribution management of multiple devices and automatically generate fault warning information when power outages, communication anomalies, or component damage are detected. Through the coordinated operation of the above structure and steps, this invention achieves integrated operation of trapping, identification, control, and remote management. Its overall technical effect cannot be obtained through simple combinations of existing technologies, demonstrating significant system-level inventiveness.

[0058] This invention belongs to the field of intelligent equipment technology for agricultural plant protection and plant quarantine, specifically relating to an intelligent trapping device for plant quarantine pests with integrated image recognition function. This device is suitable for farmland, orchards, forest areas, port quarantine zones, warehousing and logistics parks, and key ecological isolation areas. It can be used by plant quarantine departments, agricultural authorities, and large-scale planting entities for automated trapping, identification, monitoring, and remote management of quarantine pests.

[0059] In practical applications, the device of this invention can be deployed as a fixed or semi-fixed intelligent trapping terminal in the target monitoring area. Through continuous operation, it can achieve long-term dynamic monitoring of pest populations and form a regional pest monitoring network with the help of a cloud management platform, providing technical support for quarantine risk assessment and control decisions. It belongs to intelligent plant quarantine monitoring products.

[0060] like Figures 2 to 4As shown in the embodiment of the invention, each functional module works collaboratively under the unified coordination of the intelligent control module 3. During operation, the multi-band trapping light source 7 emits trapping light of the corresponding band under the control of the intelligent control module 3. Simultaneously, the specific pheromone release component 8 releases pheromones corresponding to the target pests at a set rate, creating a synergistic effect of light attraction and pheromone trapping on quarantine pests. After entering through the upper part of the trapping chamber 6, the pests fall into the pest collection trough 9 located at the bottom under the guidance of the anti-slip structure on the inner wall, achieving effective collection.

[0061] When pests enter the pest collection trough 9, the image acquisition and recognition module 2 is activated. The lens of the high-definition industrial camera 10 is pointed towards the pest collection trough 9 to capture images. The supplementary lighting component 11 starts and stops synchronously with the high-definition industrial camera 10, providing stable and uniform lighting conditions for pest imaging, ensuring clear images without obvious shadows. The acquired pest images are processed by the image recognition unit 12, which sequentially performs image preprocessing, feature extraction, and pest species and quantity identification, and outputs the identification results to the intelligent control module 3.

[0062] The intelligent control module 3 uses a microcontroller 13 as its core controller and makes logical judgments based on the recognition results fed back by the image recognition unit 12. When a quarantine pest is detected or its number reaches a preset threshold, the intelligent control module 3 triggers a local audible and visual alarm and sends a data upload command to the data transmission module 4. It also automatically adjusts the band and brightness of the multi-band trapping light source 7 and the pheromone release rate of the specific pheromone release component 8 according to the recognition results, so that the trapping parameters match the current pest type.

[0063] The data transmission module 4 uploads pest species, quantity, identification time, device location information, and operational status data to the cloud management platform via 4G / 5G, WiFi, or LoRa communication methods, and receives parameter adjustment commands from the platform, enabling bidirectional communication between the device and the platform. Simultaneously, the intelligent control module 3's built-in GPS positioning unit can collect real-time device location data. When the device experiences power outages, communication failures, or component malfunctions, it automatically generates and uploads fault warning information.

[0064] The power supply module 5 provides energy for this device. The power supply module 5 includes a solar panel, a battery, and a charging management unit. It provides stable power to each module through a complementary power supply method of solar energy and mains power, ensuring the continuous operation of the equipment in the field or remote areas.

[0065] Based on the above working principle, the embodiments of the present invention achieve the following direct technical effects: First, it realizes the integrated operation of trapping, image acquisition, identification analysis, and data transmission, reducing the degree of manual intervention; Second, by completing pest identification and judgment at the device end, it improves the real-time performance and response speed of quarantine monitoring; Third, the trapping parameters can be dynamically adjusted according to the identification results, enhancing the adaptability to different quarantine pests; Fourth, it supports remote management and multi-device networking, facilitating large-scale deployment and centralized supervision.

[0066] Example 1: Closed-loop operation implementation method for adjusting trapping parameters based on recognition result feedback

[0067] In this embodiment, the device operates continuously within the target monitoring area, collecting pests that enter the device through a trapping structure. Once the pests enter the collection area, the image acquisition unit photographs them and performs image analysis locally, outputting the pest category identification result. This identification result is transmitted directly to the control unit as input information for subsequent control decisions without manual intervention.

[0068] Based on the pest identification results, the control unit automatically adjusts the operating status of the trapping components, causing the trapping process to change according to the actual types of pests trapped. This creates a closed-loop operating mode of "trapping-identification-feedback adjustment" within the device. This operating mode can continuously complete pest monitoring and trapping control without relying on manual operation, demonstrating the overall automatic and coordinated operation capability of the device.

[0069] Example 2: Implementation of switching trapping methods based on different pest types

[0070] In this embodiment, the device classifies and identifies the trapped pests during operation, and the identification results distinguish at least different quarantine pest categories. When the identification results change, the control unit generates a corresponding trapping adjustment strategy based on different pest categories and switches the operating mode of the trapping execution structure.

[0071] In this way, the same device can employ different trapping operation states for different pest types at the same deployment location without changing the hardware structure or redeploying the equipment. This implementation method illustrates that the adjustment of the trapping execution structure is not fixed, but is driven by the recognition results, further supporting the linkage mechanism between the recognition results and the trapping behavior.

[0072] Example 3: Implementation method where the edge performs identification and directly participates in control decision-making

[0073] In this embodiment, the entire image recognition process is completed locally on the device. After acquisition, the image data is directly processed by the local recognition module to generate pest recognition results. The recognition process does not rely on remote servers or cloud computing resources, and the recognition results are generated instantly on the device.

[0074] The generated identification results are directly input to the control unit via the control interface to trigger trapping adjustments or early warning actions. Through this implementation, the device can still complete pest identification and control decisions even when the network is unstable or not connected to a remote platform, fully demonstrating the feasibility of coordinated operation between end-side identification and on-site control.

[0075] Example 4: Implementation of a Coordinated Approach to Multiple Trapping Methods

[0076] In this embodiment, the device is equipped with multiple adjustable trapping methods, all of which are managed uniformly by the control unit. During operation, the device first obtains pest identification results through the image recognition module, and then the control unit adjusts the working status of at least one trapping method based on these results.

[0077] When the monitoring environment or pest type changes, the control unit can change the currently activated trapping method or its operating parameters, enabling the device to switch between different trapping strategies under the same structural conditions. This implementation demonstrates that trapping adjustment is not a simple start-stop control, but a dynamic regulation process that corresponds to the identification results.

[0078] Example 5: Implementation method in a multi-device collaborative monitoring scenario

[0079] In this embodiment, multiple devices are deployed in the same or adjacent areas, and each device independently completes pest trapping, image recognition, and local control. While completing the recognition and control operations, each device records or transmits the recognition results and operating status information.

[0080] Centralized management of data from multiple devices enables pest distribution analysis and operational status monitoring within a specific area. This implementation demonstrates that the device can be used as an independent operating unit or as a component node in a multi-device collaborative system, supporting large-scale application scenarios.

[0081] Example 6: Implementation methods under continuous operation and abnormal conditions

[0082] In this embodiment, the device continuously performs trapping and identification operations during long-term continuous operation, and periodically updates the trapping execution status based on the identification results. When a power supply abnormality, communication abnormality, or trapping execution abnormality occurs during device operation, the control unit can identify the abnormal status and output the corresponding status information.

[0083] Even under abnormal conditions, the device can still maintain basic identification and control logic, or use the abnormal information for subsequent maintenance decisions. This implementation demonstrates that the device's identification-control mechanism is applicable not only to ideal operating conditions, but also to various operating conditions that may occur in practical applications.

[0084] The present invention provides a specific implementation scheme for an intelligent trapping device for plant quarantine pests with integrated image recognition function, as follows.

[0085] This device is deployed at a height of 1.5m in the field, with a single-point coverage radius of 50m, and uses a modular aluminum alloy waterproof shell (IP65 rating). The trapping chamber 6 of the main trapping module 1 has an inner diameter of 180mm and a height of 320mm. The inner wall is sprayed with a microstructure anti-slip coating to improve the pest retention rate. The multi-band trapping light source 7 uses a combination array of 365nm ultraviolet LEDs (5W power), 450nm blue LEDs (3W), and 850nm infrared LEDs (2W). The light intensity and band switching are controlled by PWM modulation. The ultraviolet mode is automatically activated from 18:00 to 6:00 at night, and the infrared band is superimposed from 3:00 to 5:00 in the morning to enhance phototaxis.

[0086] The specific pheromone release component 8 adopts a replaceable microporous slow-release bottle structure, with 2mL of quarantine target pest-specific pheromone inside. The release rate is controlled at 0.5mg / h. Quantitative release is achieved through a solenoid valve and a timing control module. The release time is automatically calibrated every 24 hours. When the ambient temperature exceeds 35℃, the release frequency is automatically reduced by 20%.

[0087] After the pests fall into the pest collection trough 9, a transparent observation window is installed at the bottom of the trough. The high-definition industrial camera 10 in the image acquisition and recognition module 2 uses an 8-megapixel CMOS sensor (resolution 3840×2160), a fixed focal length of 8mm, and a shooting interval of once every 10 minutes, or triggers instant shooting when the weight sensor detects an increase in mass >0.1g. The supplementary lighting component 11 uses a ring-shaped flicker-free LED light source with a color temperature of 5500K and an illuminance ≥1200lx to ensure uniform grayscale distribution in the image.

[0088] The image recognition unit 12 incorporates a convolutional neural network model based on an improved ResNet-50 architecture. The training sample library contains no fewer than 30,000 labeled images, covering eight types of quarantine pests and five types of non-target insects. The image processing workflow includes:

[0089] (1) Image denoising and histogram equalization preprocessing;

[0090] (2) Target localization based on the YOLO detection framework;

[0091] (3) Use CNN for category classification;

[0092] (4) Count the number of items using connected component analysis and the NMS algorithm;

[0093] (5) Output category confidence and quantity data.

[0094] The system's recognition accuracy reached 96.3% in actual tests, with an average recognition time of 0.38 seconds per image.

[0095] The intelligent control module 3 uses an ARM Cortex-A53 processor (1.4GHz) running an embedded Linux system, responsible for scheduling the light source, pheromone release, image acquisition, and recognition processes. The data transmission module 4 uses dual-mode communication of 4G-LTE and NB-IoT, automatically switching to NB mode when the signal strength is below -100dBm. The identified data includes: device number, timestamp, pest species code, quantity, recognition confidence level, and ambient temperature and humidity values, and is uploaded to the cloud platform database via the MQTT protocol.

[0096] The cloud system performs time-series analysis on the data and establishes an insect infestation density model:

[0097] Insect Infestation Index I = (ΣNi × Wi) / A

[0098] Where Ni represents the number of pests of type i, Wi represents the quarantine risk weight of this type of pest, and A represents the monitoring coverage area. The system issues warnings based on the I value, triggering an orange warning when I ≥ 15 and a red warning when I ≥ 30, and automatically pushes information to the monitoring terminal.

[0099] Power supply module 5 uses a combination of a 20W solar panel and a 12V 18Ah lithium battery, which can support 7 days of continuous cloudy and rainy weather. The overall daily power consumption of the system is approximately 18Wh.

[0100] Through the above-mentioned structure and data fusion, this embodiment realizes a fully automated closed loop of "trapping - collection - identification - analysis - early warning", which greatly improves the efficiency and accuracy of monitoring plant quarantine pests, reduces manual statistical errors, and realizes real-time digital management of pests.

[0101] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A smart trapping device for plant quarantine pests with integrated image recognition function, characterized in that, include: The system comprises a trapping module, an image acquisition and recognition module, an intelligent control module, and a power supply module. The main trapping module is used to physically trap and collect pests that enter the device. The image acquisition and recognition module is used to acquire images of the trapped pests and output the corresponding pest recognition results; The intelligent control module is electrically connected to the trapping main module and the image acquisition and recognition module, and is used to receive the pest recognition results and perform feedback control on the working status of the trapping main module based on the recognition results; The power supply module is used to provide working power to each module; The device employs a closed-loop collaborative mechanism consisting of "pest trapping - image recognition - control feedback," enabling the trapping behavior to automatically adjust based on the pest identification results without relying on human intervention.

2. The apparatus according to claim 1, characterized in that, The main trapping module includes a trapping chamber and a pest collection structure. The trapping chamber is a vertically connected structure, and pests enter the pest collection structure under the action of gravity to be retained.

3. The apparatus according to claim 1, characterized in that, The trapping main module also includes at least one adjustable trapping unit, the operating parameters of which are adjusted by the intelligent control module according to the pest identification results.

4. An intelligent trapping and control device for plant quarantine pests, characterized in that, include: The system includes a trapping execution unit, an image recognition unit, and a control decision-making unit. in, The trapping execution unit is used to carry out trapping operations on target pests; The image recognition unit is used to analyze images of trapped pests and output pest type identification information; The control decision unit is used to receive the pest type identification information and generate corresponding trapping and control instructions based on different pest types; The working state of the trapping execution unit is dynamically changed according to the trapping control command, thereby forming a linkage control mechanism with pest identification results as input and changes in trapping parameters as output.

5. The apparatus according to claim 4, characterized in that, The trapping control command includes at least one or more of the following: trapping intensity, trapping method, or trapping duration.

6. The apparatus according to claim 4, characterized in that, The trapping execution unit includes a photo-trapping unit and / or a chemical trapping unit.

7. An end-side intelligent identification device for monitoring plant quarantine pests, characterized in that, include: Image acquisition unit, image recognition unit, and control interface unit; The image acquisition unit is used to acquire image data of the trapped pests. The image recognition unit is used to process the image data locally and output the pest identification result; The control interface unit is used to directly output the pest identification results to an external trapping or control device; The pest identification results can be used in trapping or control decisions without relying on remote server calculations, thereby achieving coordinated operation of end-side identification and on-site control.

8. The apparatus according to claim 7, characterized in that, The image recognition unit includes a deep learning-based image classification model.

9. The apparatus according to claim 7, characterized in that, The pest identification results are output to the control interface unit within a preset time.

10. The apparatus according to claim 7, characterized in that, The intelligent control module is also used to record or transmit pest identification results and trapping and control status to support the collaborative operation of multiple devices.

Images