Intelligent tobacco insect trapping and visual monitoring device, method and system

Abstract

The invention discloses an intelligent tobacco insect trapping and visual monitoring device, method and system. The device is composed of a bottom plate and an outer cover with an intelligent dustproof grating, remote automatic opening and closing can be achieved through a grating opening and closing control module, the dustproof effect is considered through aperture differentiation and grating dislocation, meanwhile, the trapping efficiency is effectively improved, and meanwhile the problems that an existing device is prone to pollution, and the recognition precision is reduced are solved. According to the method, the system is based on multi-level AI visual analysis, the functions of tobacco insect recognition, trap plate smudginess degree evaluation and trap agent detection are integrated, and accurate insect situation recognition and plate state self-evaluation are achieved. In the system level, a communication scheduling mechanism combining three working modes of long connection, timed wake-up and LoRa remote wake-up is adopted, the requirements of real-time performance and low power consumption are both considered, and the cruising ability of equipment is remarkably improved. According to the invention, automation and intelligentization of the whole process of insect situation monitoring from trapping, identification to early warning are realized, and the efficiency and accuracy of insect pest prevention and control of tobacco storage are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of tobacco pest control, specifically to an intelligent tobacco pest trapping and visual monitoring device, method, and system. Background Technology

[0002] The constant temperature and humidity in tobacco storage and production workshops provide favorable conditions for the breeding of tobacco insects (such as tobacco beetles and tobacco mealybugs). These insects not only directly bore into raw tobacco leaves, but their excrement and eggs also contaminate tobacco products, seriously affecting cigarette quality and food safety.

[0003] Currently, common methods for controlling tobacco insects mainly include chemical fumigation, physical trapping, and manual inspection. However, these methods have significant shortcomings: while chemical fumigation is highly effective at killing insects, it poses problems such as pesticide residues, environmental pollution, and significant safety hazards, and it cannot achieve real-time monitoring and early warning of insect infestations; traditional sticky trapping relies on regular manual inspections and counting, which is inefficient, data recording is delayed, and counting errors are easily caused by visual fatigue or subjective factors. In addition, the maintenance of traps depends on manual judgment and cannot automatically identify conditions such as dirty traps or ineffective bait, leading to decreased trapping effectiveness and high maintenance costs.

[0004] Existing intelligent trapping devices, such as those disclosed in patents CN210901036U and CN219877233U, have image acquisition capabilities, but they mostly lack efficient insect identification algorithms and intelligent analysis capabilities, and cannot achieve advanced functions such as insect species classification, density assessment, and trend prediction.

[0005] Existing visual insect identification methods, such as those disclosed in patent CN110782435A, use deep learning for insect identification, but they do not fully consider the robustness issues in complex environments such as dusty workshops and changing lighting. The insect inlets and outlets cannot be opened and closed automatically, and they are not linked with the physical control of the trapping device, resulting in poor dust prevention. As dust pollution increases, the accuracy and precision of visual identification will decrease significantly.

[0006] In addition, some intelligent trapping systems, such as those disclosed in patents CN112167201A and CN113204584A, although possessing certain data analysis and early warning capabilities, mostly rely on centralized server processing, resulting in large response delays and a lack of local intelligent control and multi-mode collaborative working mechanisms.

[0007] Therefore, there is an urgent need for an intelligent tobacco pest monitoring system that integrates intelligent trapping, visual recognition, edge computing, remote control, and multi-mode communication to achieve real-time, accurate, and automated monitoring and early warning of pest infestations, thereby improving the level of intelligent pest control in the tobacco industry. Summary of the Invention

[0008] This invention addresses the various problems existing in the prior art by providing an intelligent trapping and visual monitoring device, method, and system for tobacco insects.

[0009] According to a first aspect of the present invention, a smart trapping and visual monitoring device for tobacco insects is provided, comprising:

[0010] A base plate for mounting a tobacco insect trapping board, wherein the base plate is provided with a groove for mounting the tobacco insect trapping board and a skirt mounting hole;

[0011] The outer cover is connected to the base plate. The outer cover includes a first skirt and a second skirt disposed on the opposite side of the first skirt. A through groove type first mounting hole is provided on the first skirt, and two slot type second mounting holes are provided on the second skirt. The first mounting hole and the second mounting hole are adapted to the mounting holes of the skirt.

[0012] The outer casing is equipped with a grille opening and closing control module, a vision camera module, a main control module, a communication module, and a power supply module.

[0013] The grid opening and closing control module includes a dustproof grid plate, a grid sliding module and a grid driving module, which is used to realize the remote programmable automatic sliding of the dustproof grid plate;

[0014] The visual camera module includes an embedded camera and a supplementary light, used to capture images of the tobacco insect trapping board;

[0015] The main control module is used to control the vision camera module and the grille opening and closing control module;

[0016] The communication module is connected to the main control module and includes a wireless communication module and a low-power communication module, used for data interaction with the AI ​​edge computing server;

[0017] The power module is used to supply power to the intelligent trapping and visual monitoring device for tobacco insects.

[0018] Preferably, a grid structure is provided on both sides of the outer cover to allow smoke insects to enter, and the grid hole diameter of the grid structure increases from bottom to top.

[0019] Preferably, the aperture of the dustproof grid plate increases from bottom to top, and is exactly the same as the size and arrangement of the aperture of the grid structure;

[0020] When the grid holes of the dustproof grid plate are completely aligned with the grid holes of the grid structure, the outer cover is in the open state;

[0021] When the grid holes of the dustproof grid plate partially overlap with the grid holes of the grid structure, the outer cover is in a ventilated state;

[0022] When the grid holes of the dustproof grid plate are completely misaligned with the grid holes of the grid structure, the outer cover is in a closed state.

[0023] Preferably, the grid sliding module includes sliders respectively disposed at the four corners of the outer cover, and the dustproof grid plate is driven by the sliders on both sides to achieve up and down sliding.

[0024] Preferably, the grid driving module includes an electric slide rail, a steel plate, and a limit switch. One end of the steel plate is connected to the slider, and the other end is connected to the dustproof grid plate. The electric slide rail drives the slider to move, thereby causing the dustproof grid plate to slide up and down, realizing the state switching of the outer cover.

[0025] According to a second aspect of the present invention, a method for intelligent trapping and visual monitoring of tobacco insects is provided, for use in an intelligent trapping and visual monitoring device for tobacco insects, comprising the following steps:

[0026] Images of the tobacco insect trapping board are captured using a visual camera module;

[0027] The image is enhanced to identify the number and type of tobacco insects;

[0028] The image is segmented into regions to evaluate the degree of dirtiness of the trapping board and generate a dirtiness level.

[0029] Feature extraction is performed on the image, the trapping agent is detected, and the amount of trapping agent is determined;

[0030] Based on the tobacco insect identification results, the level of soiling, and the results of attractant detection, an insect infestation early warning information is generated;

[0031] The status of the intelligent tobacco insect trapping and visual monitoring device can be monitored through the equipment management module.

[0032] The starting and stopping of the vision camera module and the sliding of the dustproof grille can be remotely controlled through the equipment control module.

[0033] Preferably, the insect infestation early warning information includes at least one of the following: an insect infestation area heat map, an insect infestation trend map, a soiling alarm, and a trap shortage alarm.

[0034] Preferably, during the insect trapping stage, the dustproof grid plate is controlled to slide downward to the bottom, so that the grid holes of the dustproof grid plate completely overlap with the grid holes of the grid structure on the side wall of the outer cover;

[0035] During the preparation stage for trapping tobacco insects, the dustproof grid plate is controlled to slide to a semi-open area, in which the grid holes of the dustproof grid plate partially overlap with the grid holes of the grid structure.

[0036] During the dust removal stage in the workshop, the dustproof grid plate is controlled to slide upward to the top, so that the grid holes of the dustproof grid plate are completely misaligned with the grid holes of the grid structure.

[0037] According to a second aspect of the present invention, an intelligent trapping and visual monitoring system for tobacco insects is provided, comprising:

[0038] Multiple intelligent trapping and visual monitoring devices for tobacco insects;

[0039] Multiple LoRa signal receiving terminals;

[0040] Network connectivity module;

[0041] Mobile software client;

[0042] AI edge computing server is used to perform functions such as tobacco insect identification, evaluation of the degree of dirtiness of the trapping board, detection of the attractant, insect infestation early warning, equipment management and equipment control.

[0043] The system supports three working modes: long connection mode, timed wake-up mode, and LoRa mode, and can dynamically switch according to network conditions and energy consumption requirements.

[0044] Preferably, at the same time, the dustproof grid plates of different intelligent insect trapping and visual monitoring devices are located in different positions.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1) Improve the automation and intelligence level of insect infestation monitoring. Achieve full automation of insect infestation identification, counting, classification, and early warning processes, reduce manual intervention, avoid subjective errors, and improve data accuracy and timeliness.

[0047] 2) Enhance system environmental adaptability and reliability. It possesses adaptive control capabilities for dustproof grid panels, adapting to dusty workshop environments and enabling pre-preparation for insect trapping; it supports multi-light source compensation and image enhancement algorithms to improve recognition robustness under complex lighting conditions.

[0048] 3) Reduce operation and maintenance costs and energy consumption. Employing multi-mode communication and power management strategies significantly reduces equipment energy consumption and extends battery life; it also supports remote maintenance and status monitoring, reducing the frequency of on-site inspections.

[0049] 4) Supports multi-level pest analysis and decision support. It features multi-dimensional analysis functions such as pest heat maps, trend analysis, dirt alarms, and bait detection, providing data support and decision-making basis for pest control.

[0050] 5) High system integration and flexible deployment. The device has a modular structure, supporting flexible adaptation of the base plate and outer cover to various specifications of trapping plates; it supports multiple communication methods such as LoRa, 5G, and WiFi, adapting to different workshop network environments. Attached Figure Description

[0051] Figure 1 This is a top view of the intelligent trapping and visual monitoring device for tobacco insects of the present invention;

[0052] Figure 2 This is a side view of the intelligent trapping and visual monitoring device for tobacco insects according to the present invention;

[0053] Figure 3 This is a top view of the slider structure of the grid sliding module of the present invention;

[0054] Figure 4 This is a flowchart of the intelligent trapping and visual monitoring method for tobacco insects according to the present invention;

[0055] Figure 5 This is a diagram of the integrated architecture of the teacher network and student network of the present invention;

[0056] Figure 6 This is a framework diagram of the intelligent trapping and visual monitoring system for tobacco insects according to the present invention. Detailed Implementation

[0057] The techniques described below can be modified in various ways and have multiple embodiments, which are described in detail below with reference to the accompanying drawings. However, this does not mean that the techniques described below are limited to the specific embodiments. It should be understood that the present invention includes all similar modifications, equivalents, and substitutions without departing from the spirit and scope of the techniques described below.

[0058] Example 1:

[0059] This embodiment provides a specific implementation of an intelligent tobacco insect trapping and visual monitoring device. For example... Figures 1 to 3 As shown, the device mainly consists of two structural parts: a base plate and an outer cover, as well as various functional components integrated inside.

[0060] See again Figure 1 The base plate is made of durable plastic or metal and features a set of three-sided positioning grooves for securing standard-sized sticky tobacco insect traps. These grooves have sufficient depth and flexibility to accommodate tobacco insect traps of varying thicknesses and sizes, thus enhancing versatility. The top and bottom of the base plate each have one and two skirt mounting holes, respectively, for connection to the outer cover.

[0061] The outer casing is a rectangular box structure with an opening at the bottom, made of opaque material to prevent external light from interfering with image acquisition. The opening of the outer casing mates with the base plate, and the interior houses a grille opening and closing control module, a vision camera module, a main control module, a communication module, and a power supply module.

[0062] The outer cover has a first skirt and a second skirt positioned opposite each other on its side edges. The first skirt has a through-slot type first mounting hole. After a mounting bolt is inserted into this first mounting hole, the bolt can move within the space defined by the first mounting hole in a direction parallel to the length of the outer cover. The second skirt has two slot type second mounting holes. After a mounting bolt is inserted into the skirt mounting hole, the bolt can slide into the second mounting hole from its side opening in a direction parallel to the length of the outer cover. The first mounting hole, the second mounting hole, and the skirt mounting hole are compatible.

[0063] When the device is installed vertically, the outer cover is secured with mounting bolts in a non-locking manner, making it easy for manual removal of the outer cover from the base plate by sliding it, so as to replace the trapping plate. When the device is installed horizontally (such as in a suspended ceiling), the outer cover is secured with mounting bolts in a locking manner to prevent the outer cover from loosening and falling off.

[0064] See again Figure 2 The outer cover has a grid structure on the left and right side walls along its length that allows smoke insects to enter. The grid holes of the grid structure increase in size from bottom to top.

[0065] Inside the outer casing, a grid opening and closing control module is further installed. This module includes a dustproof grid plate, a grid sliding module, and a grid drive module, used to achieve remote-controlled automatic sliding of the dustproof grid plate. The grid aperture of the dustproof grid plate increases in size from bottom to top, and its size and arrangement are exactly the same as those of the grid aperture in the grid structure.

[0066] See again Figure 3 The grid sliding module includes sliders located at the four corners of the outer cover. The dustproof grid plate is driven by the sliders on both sides, enabling it to slide up and down. The grid drive module includes two miniature electric slide rails, two flexible thin steel sheets, and four limit switches for positioning. One end of each steel sheet is fixed to a slider, and the other end is fixed to the dustproof grid plate. When the main control module receives a remote control command or a timed task, it controls the electric slide rails to move, pushing the sliders to move. This, in turn, drives the dustproof grid plate to slide via the steel sheets, changing the relative position of its grid holes with the grid holes of the outer cover's sidewall grid structure, thus switching the open and closed state of the outer cover. The limit switches ensure that the grid accurately stops and provides feedback when it reaches the fully open or fully closed position.

[0067] Specifically, during the smoke insect trapping stage, the dustproof grid plate is controlled to slide downwards to its lowest point, ensuring that the grid holes of the dustproof grid plate completely overlap with the grid holes of the grid structure on the side wall of the outer cover. In this state, the outer cover is fully open, and the smoke insect trapping plate on the bottom plate lures the smoke insects through the grid holes into the interior of the outer cover for capture. The grid holes gradually increase in size from bottom to top, which facilitates the smooth entry of both flying and crawling smoke insects, while minimizing the entry of floating dust, especially dust accumulated on the bottom plate and around the outer cover.

[0068] During the dust removal phase in the workshop, the dustproof grid is controlled to slide upwards to the top, ensuring that the grid holes of the dustproof grid are completely misaligned with the grid holes of the grid structure. In this state, the outer cover is completely closed, effectively preventing dust from entering.

[0069] During the preparation stage for trapping tobacco insects, the dustproof grid is slid into a semi-open area, where the grid holes of the dustproof grid partially overlap with those of the grid structure. Specifically, the relative position and degree of misalignment between the grid holes of the dustproof grid and the grid structure can be adjusted as needed. In the initial stage of preparation, the dustproof grid slowly slides downwards from the top until the small-diameter grid holes at the bottom of the dustproof grid align with the large-diameter grid holes at the top of the grid structure. At this point, the attractant odor from the tobacco insect trap can be released slowly from above, ensuring sufficient attraction time while minimizing dust accumulation. During the attraction phase of the preparation stage, the dustproof grid slides further downwards, aligning more of the grid holes of the dustproof grid with those of the grid structure, resulting in a stronger and more abundant release of the attractant odor. Throughout the preparation phase, the aperture of the interconnected grid holes decreases from the outside (at the grid structure) to the inside (at the dustproof grid plate), forming a trumpet-shaped structure. This not only allows for the release of the attractant odor and improves the trapping effect, but also effectively prevents dust from entering.

[0070] Inside the outer casing, a visual camera module is further installed, employing a low-power embedded CMOS camera module (such as the OV2640) and equipped with several adjustable-intensity LED supplementary lights. The camera lens faces the trapping plate on the base plate to ensure complete imaging of the trapping area. The main control module can control the on / off state and brightness of the supplementary lights to adapt to different ambient lighting conditions and ensure image quality.

[0071] The main control module, employing a high-performance, low-power embedded MCU (such as the STM32H7 series or a chip with an NPU, such as the Rockchip RK1808), is responsible for coordinating and controlling the entire device. Its functions include: controlling the vision camera module for timed or triggered image acquisition; controlling the motor movement of the grille drive module; preprocessing sensor data; and managing the data transmission and reception and power consumption mode switching of the communication module.

[0072] The communication module employs a multi-mode design. It includes a high-speed wireless communication module (such as a Wi-Fi module from the ESP32 series or a 5G Cat.1 module) for uploading acquired high-definition image data to the AI ​​edge computing server. It also includes a low-power wide-area network (LPWAN) communication module (such as a LoRa module using the SX1276 chip), which has extremely low power consumption and is specifically designed to receive remote wake-up signals when the device is in sleep mode.

[0073] The power module consists of multiple 18650 lithium-ion batteries connected in series and is equipped with a power management circuit (PMIC) to provide stable voltage to each module, supporting long-term field or workshop deployment.

[0074] Example 2:

[0075] This embodiment provides a specific implementation method for intelligent trapping and visual monitoring of tobacco insects. Its main process is as follows: Figure 4 As shown, it includes the following steps:

[0076] Images of the tobacco insect trapping board are captured using a visual camera module;

[0077] Image enhancement processing was performed to identify the number and type of tobacco insects;

[0078] The image is segmented into regions, the degree of dirtiness of the trapping board is evaluated, and a dirtiness level is generated;

[0079] Feature extraction is performed on the image, and the amount of bait is detected to determine the bait quantity.

[0080] Based on the tobacco insect identification results, the level of soiling, and the results of attractant detection, an insect infestation early warning information is generated;

[0081] The status of the intelligent tobacco insect trapping and visual monitoring device can be monitored through the equipment management module.

[0082] The starting and stopping of the vision camera module and the sliding of the dustproof grille can be remotely controlled through the equipment control module.

[0083] This method is applied to a smart trapping and visual monitoring device for tobacco insects, and is mainly implemented in software running on an AI edge computing server.

[0084] During the image acquisition and uploading process, the main control module of the device controls the vision camera module to acquire images of the trapping plate on the base plate according to a preset cycle (e.g., every 2 hours) or after being remotely triggered, and automatically turns on the supplementary lighting to ensure image clarity. The acquired images are uploaded to the AI ​​edge computing server via the Wi-Fi / 5G module of the communication module.

[0085] In the visual detection of insect infestations, a multi-task integrated annotation method is proposed to unify the processing of four functional tasks: insect quantity identification, insect-attracting drug missing detection, trap missing detection, and dirt level judgment. This method unifies the above four functional tasks into three logical tasks: (1) tobacco insect target detection; (2) insect-attracting drug target detection; and (3) trap target detection. By unifying quantity statistics, missing detection, and dirt level classification into target detection tasks, an end-to-end detection network is constructed, thereby avoiding the redundancy and inefficiency problems caused by multi-model collaboration.

[0086] In the implementation, a unified rectangular bounding box annotation method is used to locate and classify tobacco beetles, insect attractants, and traps in the image. Tobacco beetles are further subdivided into tobacco beetles (category 0) and tobacco mealybugs (category 2) for easier subsequent quantity counting and species identification; insect attractants are uniformly labeled as category 1, used only to determine their presence; traps are directly labeled according to their level of soiling (0-4) to simultaneously achieve missing data detection and soiling degree assessment. This annotation method not only simplifies the data annotation process but also facilitates the model's simultaneous learning of multiple tasks within a single network.

[0087] To enhance the model's adaptability and generalization ability across different tobacco factory scenarios, this invention further designs a cross-domain progressive learning mechanism based on teacher-student knowledge distillation, see [link to relevant documentation]. Figure 5 This mechanism uses YOLOv8 as the base network, constructing identical teacher and student networks. The student network's detection head is expanded with new category output nodes to adapt to incremental tasks. During training, the teacher network loads and freezes the weights from the old scene, while the student network updates its weights based on the new scene data. Simultaneously, a cross-head connection mechanism is used to calculate the distillation loss using the cross-prediction results generated by the teacher network based on student features. This allows the student network to retain its ability to recognize old tasks while learning new task knowledge, effectively avoiding catastrophic forgetting.

[0088] Specifically, the loss function used for backpropagation of the branches trained with the new data labels is denoted as:

[0089]

[0090] in, This represents the loss during training with new data labels; This indicates the number of positive samples, that is, the number of samples in the image that contain the target; The classification loss is represented by Quality Focal Loss, which measures the difference between the classification predictions of the student network and the classification labels of the teacher network. Represents the classification prediction of the student network; It is a specified function. Specifically, if there is a visible foreign object in the prediction box, the function will return 1; otherwise, it will return 0. and These are hyperparameters used to balance the importance of different loss terms; The regression loss is represented by GIOU Loss, which measures the difference between the regression predictions of the student network and the regression labels of the teacher network. Represents the bounding box prediction of the student network; The label of the detection box indicating the teacher network; It stands for Distribution Focal Loss, which guides the network to quickly focus on the distribution in the vicinity of the target location, where z represents the number of samples.

[0091] The branches guided by the old task network to the new task network use distillation loss, defined as follows:

[0092]

[0093] in, This represents the distillation loss function guided by the old scenario model, used to measure the difference between the student model and the teacher model during the knowledge distillation process. The distillation loss function represents the classification task and is used to measure the difference between the student model and the teacher model on the classification task. This indicates the classification result of cross-prediction; This represents the classification results of the teacher model; The distillation loss function represents the detection task and is used to measure the difference between the student model and the teacher model on the detection task. This represents the prediction results of the student model on the detection task; This represents the prediction results of the teacher model on the detection task, including bounding boxes and confidence scores.

[0094] During the prediction phase, the output of the logical task can be directly mapped to the results of the functional tasks: the tobacco insect detection results are used to count the number and distribution of pests; a detection count of 0 for insect attractants is considered missing; a detection count of 0 for traps is also considered missing; and other output categories (1-3) serve as the basis for determining the degree of soiling. This method combines multi-task integrated annotation with a cross-domain progressive learning mechanism, effectively integrating multiple functional tasks, significantly improving detection efficiency and system real-time performance, and is suitable for large-scale, multi-scenario pest monitoring needs.

[0095] In the pest infestation early warning and decision support stage, the pest infestation early warning module integrates the results of the above three logical tasks. On the one hand, it generates a heat map of the infestation area, combining the identification results of multiple devices with their geographical location information to visualize the distribution of pest density on the floor plan of the workshop or warehouse in the form of a heat map. On the other hand, it generates a pest infestation trend map, analyzing the changing trend of pest population in a specific area based on historical data to make short-term predictions. In addition, it triggers alarms. When the pest population exceeds a threshold, or when an alarm is received regarding missing / dirty / missing bait traps, an early warning message is sent to management personnel through the main control system software or mobile terminal, prompting them to take targeted measures.

[0096] In the equipment management and control phase, the equipment management module monitors the operating status of each device in real time (power level, network signal, grille status, etc.). The equipment control module allows users to remotely and manually control the image acquisition, grille opening and closing, and other actions of individual or batch devices.

[0097] Example 3:

[0098] This embodiment provides a specific implementation of an intelligent tobacco insect trapping and visual monitoring system. Its main framework is as follows: Figure 6 As shown, it includes:

[0099] Multiple intelligent trapping and visual monitoring devices for tobacco insects;

[0100] Multiple LoRa signal receiving terminals;

[0101] Network connectivity module;

[0102] Mobile software client;

[0103] The AI ​​edge computing server is used to perform functions such as tobacco insect identification, evaluation of the degree of soiling of the trapping board, detection of the attractant, insect infestation early warning, equipment management and equipment control.

[0104] The system adopts a collaborative "end-edge-cloud" architecture. The end-user layer consists of multiple intelligent insect trapping and visual monitoring devices, as described in Example 1, deployed in tobacco warehouses or workshops. These devices are responsible for insect trapping, on-site data collection, and execution of relevant commands.

[0105] The network layer includes local area networks (Wi-Fi), cellular networks (5G / 4G), and low-power wide area networks (LoRa). The LoRa gateway is responsible for receiving wake-up signals and a small amount of status data, while Wi-Fi / 5G is responsible for high-bandwidth image data transmission.

[0106] For the edge layer, one or more AI edge computing servers are deployed within the factory area. These servers run all the AI ​​algorithms and functional modules described in Example 2 (tobacco insect identification, dirt assessment, attractant detection, insect infestation early warning, equipment management, and equipment control). This edge computing model reduces data transmission latency to the cloud, improving system response speed and data privacy.

[0107] At the application layer, users can access the system through system software (Web-based control panel) or a mobile app. The interface can display real-time pest data, heat maps, trend analysis, equipment status, alarm lists, etc., and provides a remote control entry point.

[0108] The system supports three working modes: long connection mode, timed wake-up mode, and LoRa mode, and can dynamically switch according to network conditions and energy consumption requirements.

[0109] Long-term connection mode: Suitable for scenarios with grid power supply and high real-time requirements. The device maintains a persistent connection with the AI ​​edge computing server via Wi-Fi / 5G, enabling near real-time image transmission and command control.

[0110] Timed wake-up mode: Designed to balance power consumption and responsiveness. Users can set the heartbeat interval T (e.g., 6 hours). The device is in sleep mode most of the time, only keeping the RTC clock running. It wakes up every T interval to actively connect to the server to check if there are any instructions to be executed (e.g., take a picture immediately), and then goes back to sleep after execution.

[0111] LoRa Mode (Ultra-Low Power Mode): Suitable for battery-powered scenarios requiring long battery life. Normally, only the LoRa module is in an extremely low-power listening state. When the user needs to issue commands via the mobile or main control system software, the mobile or main control system software first notifies the LoRa gateway (LoraAP). The gateway then wakes up all smart insect trapping and visual monitoring devices via LoRa network broadcast. After being woken up, the device immediately activates its Wi-Fi / 5G module, connects to the AI ​​edge computing server to obtain detailed instructions, and executes them. This mode can significantly extend battery life.

[0112] In practical use, the dustproof grid panels of different intelligent tobacco insect trapping and visual monitoring devices can be positioned in the same or different places at the same time, thus flexibly adapting to different usage scenarios and achieving a reasonable balance between cleaning and dust prevention and tobacco insect trapping. For example, during the tobacco insect trapping stage, for devices located on the periphery of the area, the grid holes of the dustproof grid panel are partially misaligned and partially overlap with the grid holes of the grid structure; for devices located inside the area, the grid holes of the dustproof grid panel are completely overlapped with the grid holes of the grid structure.

[0113] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a readable storage medium, at least one input device, and at least one output device, and transmitting data and instructions to the readable storage medium, the at least one input device, and the at least one output device.

[0114] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0115] In the context of this invention, a readable storage medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A readable storage medium can be a machine-readable signal medium or a machine-readable storage medium. A readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0116] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, as some steps can be performed in other orders or simultaneously according to the present invention. Secondly, those skilled in the art should also understand that the embodiments described in the specification are optional embodiments, and the actions and modules involved are not necessarily essential to the present invention. It should be understood that the various forms of processes described above can be used, with steps reordered, added, or deleted. For example, the steps described in the present invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of the present invention can be achieved, and this is not limited herein. The above specific embodiments do not constitute a limitation on the scope of protection of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart trapping and visual monitoring device for tobacco insects, characterized in that, include: A base plate for mounting a tobacco insect trapping board, wherein the base plate is provided with a groove for mounting the tobacco insect trapping board and a skirt mounting hole; The outer cover is connected to the base plate. The outer cover includes a first skirt and a second skirt disposed on the opposite side of the first skirt. A through groove type first mounting hole is provided on the first skirt, and two slot type second mounting holes are provided on the second skirt. The first mounting hole and the second mounting hole are adapted to the mounting holes of the skirt. The outer casing is equipped with a grille opening and closing control module, a vision camera module, a main control module, a communication module, and a power supply module. The grid opening and closing control module includes a dustproof grid plate, a grid sliding module and a grid driving module, which is used to realize the remote programmable automatic sliding of the dustproof grid plate; The visual camera module includes an embedded camera and a supplementary light, used to capture images of the tobacco insect trapping board; The main control module is used to control the vision camera module and the grille opening and closing control module; The communication module is connected to the main control module and includes a wireless communication module and a low-power communication module, used for data interaction with the AI ​​edge computing server; The power module is used to supply power to the intelligent trapping and visual monitoring device for tobacco insects.

2. The apparatus as described in claim 1, characterized in that, The outer cover has grid structures on both sides that allow smoke insects to enter, and the grid holes of the grid structures increase in size from bottom to top.

3. The apparatus as described in claim 2, characterized in that, The diameter of the grid holes in the dustproof grid plate increases from bottom to top, and is exactly the same as the size and arrangement of the grid holes in the grid structure. When the grid holes of the dustproof grid plate are completely aligned with the grid holes of the grid structure, the outer cover is in the open state; When the grid holes of the dustproof grid plate partially overlap with the grid holes of the grid structure, the outer cover is in a ventilated state; When the grid holes of the dustproof grid plate are completely misaligned with the grid holes of the grid structure, the outer cover is in a closed state.

4. The apparatus as described in claim 3, characterized in that, The grid sliding module includes sliders respectively disposed at the four corners of the outer cover. The dustproof grid plate is driven by the sliders on both sides, thereby realizing up and down sliding.

5. The apparatus as described in claim 4, characterized in that, The grid driving module includes an electric slide rail, a steel plate, and a limit switch. One end of the steel plate is connected to the slider, and the other end is connected to the dustproof grid plate. The electric slide rail drives the slider to move, thereby causing the dustproof grid plate to slide up and down, realizing the state switching of the outer cover.

6. A method for intelligent trapping and visual monitoring of tobacco insects, used in the intelligent trapping and visual monitoring device for tobacco insects as described in any one of claims 1-5, characterized in that, Includes the following steps: Images of the tobacco insect trapping board are captured using a visual camera module; The image is enhanced to identify the number and type of tobacco insects; The image is segmented into regions to evaluate the degree of dirtiness of the trapping board and generate a dirtiness level. Feature extraction is performed on the image, the trapping agent is detected, and the amount of trapping agent is determined; Based on the tobacco insect identification results, the level of soiling, and the results of attractant detection, an insect infestation early warning information is generated; The status of the intelligent tobacco insect trapping and visual monitoring device can be monitored through the equipment management module. The starting and stopping of the vision camera module and the sliding of the dustproof grille can be remotely controlled through the equipment control module.

7. The method as described in claim 6, characterized in that, The pest infestation early warning information includes at least one of the following: a heat map of the infestation area, a trend map of pest changes, a dirt alarm, and a trap shortage alarm.

8. The method as described in claim 7, characterized in that, During the tobacco insect trapping stage, the dustproof grid plate is controlled to slide down to the bottom, so that the grid holes of the dustproof grid plate completely overlap with the grid holes of the grid structure on the side wall of the outer cover. During the preparation stage for trapping tobacco insects, the dustproof grid plate is controlled to slide to a semi-open area, in which the grid holes of the dustproof grid plate partially overlap with the grid holes of the grid structure. During the dust removal stage in the workshop, the dustproof grid plate is controlled to slide upward to the top, so that the grid holes of the dustproof grid plate are completely misaligned with the grid holes of the grid structure.

9. A smart trapping and visual monitoring system for tobacco insects, characterized in that, include: Multiple intelligent trapping and visual monitoring devices for tobacco insects as described in any one of claims 1-5; Multiple LoRa signal receiving terminals; Network connectivity module; Mobile software client; AI edge computing server is used to perform functions such as tobacco insect identification, evaluation of the degree of dirtiness of the trapping board, detection of the attractant, insect infestation early warning, equipment management and equipment control. The system supports three working modes: long connection mode, timed wake-up mode, and LoRa mode, and can dynamically switch according to network conditions and energy consumption requirements.

10. The system as described in claim 9, characterized in that, At the same time, the dustproof grid plates of different intelligent trapping and visual monitoring devices for tobacco insects are located in different positions.

Images