A corn disease and pest intelligent identification and prevention device based on machine vision

Abstract

This invention provides a machine vision-based intelligent identification and control device for corn diseases and pests, relating to the field of intelligent control of corn diseases and pests. The machine vision-based intelligent identification and control device for corn diseases and pests includes: an image acquisition module for acquiring images of corn plants; and an image processing module for preprocessing the acquired images. The machine vision-based intelligent identification and control device for corn diseases and pests provided by this invention utilizes the image acquisition module to automatically acquire field images, uses the image processing module to process the acquired images for more accurate subsequent identification and analysis, and then uses the identification and analysis module to intelligently identify diseases and pests. Based on the identified disease and pest categories, the execution module performs control operations such as pesticide application on the areas of plants with diseases and pests, thus meeting the development needs of intelligent, green, and economical agriculture.

Description

Technical Field

[0001] This invention relates to the field of intelligent prevention and control of corn diseases and pests, and in particular to an intelligent identification and control device for corn diseases and pests based on machine vision. Background Technology

[0002] As a major food and cash crop, corn has a wide planting area and a large industrial scale, playing an important role in agricultural food security and rural economic development. Throughout its growth cycle, corn is susceptible to various diseases and pests, such as large leaf spot, small leaf spot, rust, aphids, corn borers, and fall armyworms. If monitoring is not timely and control measures are delayed, it can easily lead to stunted plant growth and a significant reduction in yield, seriously affecting corn quality and the economic benefits of planting.

[0003] Current technologies for controlling corn diseases and pests largely rely on manual field inspections, depending on growers' experience and visual assessment of pest types and severity. This approach has significant drawbacks: manual inspections are inefficient, time-consuming, labor-intensive, and heavily influenced by subjective experience, leading to missed detections and misjudgments. Furthermore, traditional control methods often employ extensive, uniform spraying of pesticides over large areas, lacking targeted application, resulting in wasted pesticide resources, increased planting costs, and potential ecological and food safety issues such as soil and water pollution. Additionally, the lack of systematic data collection, storage, and traceability for field pests and diseases hinders the analysis of pest and disease occurrence patterns, early warning systems, and scientific management. Therefore, there is still room for research into how to combine AI with images of corn growth for intelligent control.

[0004] Therefore, it is necessary to provide a machine vision-based intelligent identification and control device for corn diseases and pests to solve the above-mentioned technical problems. Summary of the Invention

[0005] This invention provides a machine vision-based intelligent identification and control device for corn diseases and pests, which solves the problem of how to combine AI with images of corn growth for intelligent control.

[0006] To solve the above-mentioned technical problems, the present invention provides a machine vision-based intelligent identification and control device for corn diseases and pests, comprising:

[0007] Image acquisition module, the image acquisition module being used to acquire images of corn plants;

[0008] An image processing module is used to preprocess the acquired images; the preprocessing includes size normalization, noise reduction, illumination and color correction, image sharpening, pixel normalization, and invalid image filtering.

[0009] A recognition and analysis module is constructed based on a deep learning model. This module is used to identify and analyze the collected maize plant images and generate prevention and control plans.

[0010] The execution module is used to implement corresponding prevention and control treatments on corn plants in areas with pests and diseases;

[0011] The data management module is used to store and manage preprocessed images, pest and disease identification results, and field control operation records.

[0012] Preferably, the method for constructing the identification and analysis module specifically includes the following steps:

[0013] S1. Collect images of corn at different growth stages, including seedling stage, jointing stage, and ear stage, including healthy plants and plants with pests and diseases.

[0014] S2. Label, preprocess, and augment the acquired image data;

[0015] S3. Divide the processed image data into a dataset, including a training set, a validation set, and a test set;

[0016] S4. Select a lightweight deep learning object detection model as the basic framework, input the divided training set samples into the lightweight deep learning object detection model, and perform iterative training.

[0017] S5. Use the test set to test the trained model;

[0018] S6. Deploy the trained model to the corn disease and pest identification and control system.

[0019] Preferably, the image acquisition module includes a fixed acquisition unit and a mobile acquisition unit. The fixed acquisition units are distributed in multiple locations in the field. The fixed acquisition unit includes a mounting box, a lifting device, and a detection component. The lifting device is mounted on the mounting box. The detection component includes a mounting plate and a detection camera. The mounting plate is mounted on the output end of the lifting device. The detection camera is mounted on the mounting plate via a pitch and rotation component.

[0020] Preferably, the lifting device includes a drive motor, a lead screw, a nut, a U-shaped sleeve, and a positioning rod. The drive motor is installed inside the mounting box. The bottom end of the lead screw is fixedly connected to the drive shaft of the drive motor. The nut is threaded onto the lead screw. The positioning rod is arranged adjacent to the lead screw. The nut is sleeved on the positioning rod through the U-shaped sleeve. The mounting plate is installed on the nut.

[0021] Preferably, an installation cylinder is rotatably mounted on the top of the installation box. The installation cylinder is sleeved outside the lead screw and the positioning rod. The top of the lead screw is rotatably connected to the installation cylinder. Both sides of the installation cylinder have slotted openings. One end of the installation plate passes through one of the slotted openings and is fixedly connected to the nut.

[0022] The machine vision-based intelligent identification and control device for corn diseases and pests also includes a drive component, which is used to drive the mounting cylinder to rotate.

[0023] Preferably, the machine vision-based intelligent identification and control device for corn diseases and pests further includes a cleaning component, which includes a storage box, a cleaning brush, a rotating shaft, and a rotating component. The storage box is mounted on the mounting cylinder, one end of the cleaning brush handle is rotatably mounted on the storage box via the rotating shaft, and the rotating component is used to drive the cleaning brush to rotate.

[0024] Preferably, it also includes a driving device and a transmission component;

[0025] The rotating component is a first gear, which is fixed to one end of the rotating shaft. The bottom end of the positioning rod is slidably connected to the top of the mounting box via a slider.

[0026] The pushing device includes a push cylinder and an L-shaped frame. The push cylinder is mounted on the mounting box. The output end of the push cylinder passes through the side wall of the mounting box and is connected to the L-shaped frame. The top end of the L-shaped frame passes through the top of the mounting box through a sliding hole and is connected to the slider.

[0027] The transmission component includes a connecting rod and a first toothed plate. One end of the connecting rod passes through the other side wall of the mounting box and is connected to the L-shaped frame. The first toothed plate is installed at the other end of the connecting rod and is aligned with the lower side of the first gear.

[0028] The driving component includes a sleeve, a connecting shaft, and a connector. The sleeve is fixedly installed in the mounting cylinder. One end of the connecting shaft is slidably connected inside the sleeve, and the other end is aligned with an insertion hole on the drive shaft. The connector connects the connecting shaft and the slider.

[0029] Preferably, the connector is a connecting ring, the sleeve has strip holes on both sides, the connecting shaft has a rectangular through hole, the connector is sleeved on the outside of the drive shaft, one end is fixedly connected to the slider, and the other end passes through the strip hole and the rectangular through hole;

[0030] When the slider drives the connecting shaft to insert into the socket to a preset depth via the connector, the center of the connector coincides with the center of the drive shaft.

[0031] Preferably, a positioning groove is provided on one side of the inner wall of the rectangular through hole, and a protrusion is provided on one side of the connector to be inserted into the positioning groove.

[0032] Preferably, the cleaning brush includes an assembly rod, a connecting plate, a one-way gear, a second toothed plate, and a plurality of cleaning sponges. The connecting plate is horizontally mounted on the brush handle. The assembly rod passes through the connecting plate and is rotatably connected to the connecting plate. The plurality of cleaning sponges are evenly mounted on the assembly rod circumferentially. The one-way gear is mounted on the bottom end of the assembly rod. The second toothed plate is mounted on the connecting rod and is spaced apart from the first toothed plate.

[0033] When the rotating shaft drives the cleaning brush to rotate from horizontal to vertical, the second toothed plate aligns with the one-way gear.

[0034] Compared with related technologies, the intelligent identification and control device for corn diseases and pests based on machine vision provided by this invention has the following beneficial effects:

[0035] This invention provides a machine vision-based intelligent identification and control device for corn diseases and pests. By combining machine vision image acquisition technology with a lightweight deep learning target detection model, an intelligent identification and control system for corn diseases and pests is built. The image acquisition module automatically collects field images, and the image processing module processes the collected images to facilitate more accurate identification and analysis. Then, the identification and analysis module intelligently identifies diseases and pests. Based on the identified disease and pest categories, the execution module performs control operations such as pesticide application on the affected plant areas, thus meeting the development needs of intelligent, green, and economical agriculture. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the principle block structure of the intelligent identification and control device for corn diseases and pests based on machine vision provided by the present invention.

[0037] Figure 2 This is a schematic diagram of the structure of the fixed acquisition unit provided by the present invention;

[0038] Figure 3 for Figure 2 A cross-sectional view of the fixed acquisition unit shown;

[0039] Figure 4 for Figure 3 The enlarged schematic diagram of part A shown below;

[0040] Figure 5 This is a partial structural schematic diagram of the fixed acquisition unit provided by the present invention;

[0041] Figure 6 A schematic diagram showing the cleaning brush of the present invention in operation;

[0042] Figure 7 This is a schematic diagram illustrating the steps of inserting a connecting shaft into a socket according to the present invention, wherein... Figure 7 (a) is a schematic diagram showing the insert shaft not inserted into the socket. Figure 7 Image (b) is a schematic diagram showing the separation of the protrusion and the positioning groove. Figure 7 (c) is a schematic diagram of the insertion of the insert shaft into the insertion hole;

[0043] Figure 8 This is a schematic diagram of another embodiment of the cleaning brush provided by the present invention, wherein, Figure 8 (a) is a schematic diagram showing the state of the cleaning brush being collected in the storage box. Figure 8 (b) is a schematic diagram showing the cleaning brush rotated to a vertical position. Figure 8 (c) is a schematic diagram showing the state in which the second gear plate drives the one-way gear to rotate the assembly rod.

[0044] Numbering on the map:

[0045] 1. Mounting box; 101. Sliding hole;

[0046] 2. Drive motor; 21. Drive shaft; 211. Socket;

[0047] 3. Lifting device; 31. Lead screw; 32. Nut; 33. U-shaped sleeve; 34. Positioning rod; 35. Sliding block;

[0048] 4. Detection components; 41. Mounting plate; 42. Detection camera; 43. Protective cover;

[0049] 5. Mounting cylinder; 501. Strip-shaped opening;

[0050] 6. Pushing device; 61. Push cylinder; 62. L-shaped frame;

[0051] 7. Sweeping component; 71. Storage box; 72. Sweeping brush; 73. Rotating shaft; 74. First gear;

[0052] 721. Brush handle; 722. Cleaning sponge;

[0053] 8. Transmission component; 81. Connecting rod; 82. First gear plate;

[0054] 9. Drive component; 91. Sleeve; 92. Connecting shaft; 93. Connecting component;

[0055] 921. Rectangular through hole; 922. Positioning groove; 931. Protrusion;

[0056] 723. Assembly rod; 724. Connecting plate; 725. One-way gear; 83. Second gear plate. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0058] This invention provides a machine vision-based intelligent identification and control device for corn diseases and pests.

[0059] First Embodiment

[0060] Please refer to the following: Figure 1 In one embodiment of the present invention, the machine vision-based intelligent identification and control device for corn diseases and pests includes:

[0061] Image acquisition module, the image acquisition module being used to acquire images of corn plants;

[0062] An image processing module is provided, which is used to preprocess the acquired images. The preprocessing includes size normalization, noise reduction, illumination and color correction, image sharpening, pixel normalization, and invalid image filtering.

[0063] A recognition and analysis module is constructed based on a deep learning model. This module is used to identify and analyze the collected maize plant images and generate prevention and control plans.

[0064] The execution module is used to implement corresponding prevention and control treatments on corn plants in areas with pests and diseases;

[0065] The data management module is used to store and manage preprocessed images, pest and disease identification results, and field control operation records.

[0066] By combining machine vision image acquisition technology with a lightweight deep learning object detection model, an intelligent identification and control system for corn diseases and pests is built. The image acquisition module automatically collects field images, and the image processing module processes the collected images to facilitate more accurate identification and analysis. Then, the identification and analysis module intelligently identifies diseases and pests. Based on the identified disease and pest categories, the execution module performs control operations such as pesticide application on the affected plant areas, thus meeting the development needs of intelligent, green, and economical agriculture.

[0067] Staff can use the data management module to search for collected images and historical data.

[0068] During image processing, field camera images may contain a lot of useless information such as sky, land, field ridges, and weeds. Therefore, only the corn plant leaf area is selected and preserved, and invalid background is cropped out to reduce computational load and avoid interfering with model recognition.

[0069] Image size normalization is necessary because deep learning models typically require a fixed input size, commonly 640×640. This involves scaling up images of varying sizes captured on-site to the size specified by the model.

[0070] Image denoising is crucial in field photography, which often includes dust noise, nighttime particle noise, and slight water mist blur. Denoising can be achieved through filtering (Gaussian filtering, mean filtering), smoothing the image, preserving the edges of blemishes, and removing interfering noise.

[0071] Light and color correction are necessary because field lighting conditions vary greatly: direct sunlight, cloudy days, shade, and backlighting. This requires: brightness balancing, contrast enhancement, and white balance correction to restore normal leaf and lesion colors and prevent model misidentification or omissions due to light and dark discrepancies.

[0072] Image sharpening is crucial because long-distance shots taken in the field tend to be blurry. Edge sharpening enhances the outlines of lesions and insect textures, improving the accuracy of subsequent deep learning recognition.

[0073] Pixel normalization scales image pixel values ​​from 0 to 255 to the range of 0 to 1, conforming to the input data specifications of deep learning models, thus accelerating inference and ensuring stable recognition.

[0074] Invalid image filtering and automatic rejection: Overexposed, completely black, severely occluded, and lens-contaminated images are not sent to the model for recognition, saving equipment computing power and avoiding invalid calculations and incorrect recognition.

[0075] The execution modules include spraying drones, trapping equipment (such as pheromone lures and insecticidal lamps) or robotic spraying equipment, and spraying equipment deployed in the field.

[0076] In this embodiment, the method for constructing the identification and analysis module specifically includes the following steps:

[0077] S1. Collect images of corn at different growth stages, including seedling stage, jointing stage, and ear stage, including healthy plants and plants with pests and diseases.

[0078] S2. Label, preprocess, and augment the acquired image data;

[0079] S3. Divide the processed image data into a dataset, including a training set, a validation set, and a test set;

[0080] S4. Select a lightweight deep learning object detection model as the basic framework, input the divided training set samples into the lightweight deep learning object detection model, and perform iterative training.

[0081] S5. Use the test set to test the trained model;

[0082] S6. Deploy the trained model to the corn disease and pest identification and control system.

[0083] In S1, corn diseases and pests include common categories such as corn leaf blight, small leaf blight, rust, corn borer, aphids, and fall armyworm; at the same time, it can supplement the publicly available agricultural dataset samples to form an original image library, ensuring that the samples cover complex field scenarios such as different lighting, weather, and shading.

[0084] In S2, the LabelImg annotation tool is used to select and annotate target areas such as lesions, insects, and damaged stems and leaves in the image, defining specific labels for various pests and diseases. After annotation, corresponding annotation files are generated to establish the image-label correspondence and provide standard samples for model training. Images are uniformly cropped and scaled to a fixed size (e.g., 640×640). Image pixels are normalized to reduce data difference interference. Data augmentation is performed through rotation, left and right flipping, brightness fine-tuning, contrast transformation, and random noise addition to expand the number of samples and improve the model's adaptability to complex field environments.

[0085] In S3, the training set, validation set, and test set are typically in a 7:2:1 ratio. The training set is used for the model to learn features, the validation set is used for parameter tuning during training, and the test set is used for final model performance evaluation.

[0086] S4. YOLOv8 (You Only Look Once Version 8, a lightweight object detection model based on convolutional neural networks) is selected as the basic framework. It relies on convolutional neural networks to autonomously extract deep features such as texture, contour, and color of pests and diseases. The training environment is set up and hyperparameters such as batch_size, number of iterations, and learning rate are configured to adapt to the subsequent embedded deployment requirements of field equipment.

[0087] In S5, the pre-divided dataset is input into the network model, and iterative training is performed using GPU. The model extracts pest and disease features through forward propagation, and then continuously corrects the network weights and reduces the prediction error through backpropagation. It continues to iterate until the model accuracy and loss value tend to stabilize, thus completing the autonomous feature learning.

[0088] The trained model was tested using a test set, and core metrics such as precision, recall, and mAP were calculated. To address issues such as missed detections, false detections, and poor recognition of small targets, the model was iteratively optimized by supplementing samples, adjusting hyperparameters, and optimizing the network structure. Ultimately, a deep learning model for identifying corn diseases and pests that is suitable for field scenarios was obtained.

[0089] Please see Figure 2 and Figure 3 In this embodiment, the image acquisition module includes a fixed acquisition unit and a mobile acquisition unit. The fixed acquisition unit is distributed in multiple locations in the field. The fixed acquisition unit includes a mounting box 1, a lifting device 3, and a detection component 4. The lifting device 3 is mounted on the mounting box 1. The detection component 4 includes a mounting plate 41 and a detection camera 42. The mounting plate 41 is mounted on the output end of the lifting device 3. The detection camera 42 is mounted on the mounting plate 41 via a pitch and rotation component.

[0090] The mobile data acquisition unit includes drone data acquisition equipment, that is, a drone carrying a detection camera 42 to detect any area, or a walking robot carrying a detection camera 42 to detect.

[0091] Fixed acquisition units are used to be fixedly installed on field ridges or inside the field to take pictures of corn plants at preset intervals (such as hourly / daily) or continuously; each fixed acquisition unit detects a corresponding small area.

[0092] By setting up the lifting device 3 and adjusting the height of the detection component 4, it is possible to adapt to different heights at different growth stages of corn for image acquisition, and to collect images from different locations of the corn plant according to the image acquisition requirements, such as the stems, leaves, corn fruits, or corn silks at different locations of the plant.

[0093] Furthermore, the detection camera 42 is mounted on the mounting plate 41 via a pitch rotator, allowing for pitch angle adjustment and better control of the observation angle.

[0094] The pitch and rotation component includes two assembly blocks, a rotating shaft, and a rotating component. The two assembly blocks are symmetrically mounted on the mounting plate 41. The two ends of the rotating shaft pass through the two assembly blocks and are rotatably connected to them. The detection camera 42 is fixedly mounted on the rotating shaft by a fastener. The rotating component can be a rotary motor or a positioning shaft. The rotary motor is mounted on the mounting plate 41 by a mounting bracket. The output shaft of the rotary motor is fixedly connected to the rotating shaft. The angle of the detection camera 42 is adjusted by the rotary motor.

[0095] When used as a positioning shaft, the positioning shaft passes through the assembly block and is threadedly connected to the assembly block. After adjusting the angle of the detection camera 42, the positioning shaft is turned to press against the detection camera 42 to achieve positioning.

[0096] A fill light (not shown) is installed on the mounting plate 41 to facilitate image acquisition at night.

[0097] Preferably, the detection component 4 also includes a protective cover 43, which is mounted on the mounting plate 41 by a fixing rod. The protective cover 43 can protect and shield the detection camera 42, preventing rainwater from falling directly onto the detection camera 42.

[0098] Please see Figure 5 In this embodiment, the lifting device 3 includes a drive motor 2, a lead screw 31, a nut 32, a U-shaped sleeve 33, and a positioning rod 34. The drive motor 2 is installed in the mounting box 1. The bottom end of the lead screw 31 is fixedly connected to the drive shaft 21 of the drive motor 2. The nut 32 is threadedly connected to the lead screw 31. The positioning rod 34 is arranged adjacent to the lead screw 31. The nut 32 is sleeved on the positioning rod 34 through the U-shaped sleeve 33. The mounting plate 41 is installed on the nut 32.

[0099] When it is necessary to adjust the detection camera 42, the drive motor 2 drives the lead screw 31 to rotate clockwise or counterclockwise, so that the nut 32 moves up or down along the lead screw 31 and the positioning rod 34. The nut 32 drives the detection camera 42 to move up or down through the mounting plate 41, thereby realizing the function of raising and lowering the detection camera 42 in height.

[0100] In other embodiments, the lifting device 3 may also use a chain lifting structure or a winch lifting structure. When using a chain or rope for lifting, a positioning rod 34 is still provided to limit the mounting plate 41 to ensure the stability of the detection camera 42.

[0101] The wiring connected to the detection camera 42 meets the spatial distance requirements for adjusting the position of the detection camera 42, such as raising or lowering it.

[0102] Please see Figure 2 and Figure 3 The top of the mounting box 1 is rotatably mounted with a mounting cylinder 5. The mounting cylinder 5 is sleeved on the outside of the lead screw 31 and the positioning rod 34. The top of the lead screw 31 is rotatably connected to the mounting cylinder 5. Both sides of the mounting cylinder 5 are provided with strip-shaped openings 501. One end of the mounting plate 41 passes through one of the strip-shaped openings 501 and is fixedly connected to the nut 32.

[0103] The machine vision-based intelligent identification and control device for corn diseases and pests also includes a drive component 9, which is used to drive the mounting cylinder 5 to rotate.

[0104] The mounting plate 41 has a connecting part on one side. The width of the connecting part is adapted to the width of the slot 501. The connecting part passes through the slot 501 and is fixedly connected to the nut 32.

[0105] The installation sleeve 5 is used to protect the lead screw 31 and the positioning rod 34, preventing them from being directly exposed to the outside and extending their service life.

[0106] Furthermore, the mounting cylinder 5 is driven to rotate by the drive component 9, thereby rotating the mounting cylinder 5 with the drive mounting plate 41 via the strip-shaped opening 501, which in turn drives the detection camera 42 to rotate circumferentially. This allows the horizontal angle of the detection camera 42 to be adjusted, enabling the detection of corn plants in different areas as needed. Moreover, by directly driving the mounting cylinder 5 to rotate, the horizontal angle of the detection camera 42 can be adjusted over a wider range. However, when directly adjusting the angle of the detection camera 42, the acquisition range will be blocked by the mounting cylinder 5.

[0107] A photovoltaic panel can be installed on the top of the mounting cylinder 5 to provide additional power.

[0108] As an optional embodiment, the drive component 9 includes a drive motor, a main gear, and a driven gear. The drive motor is mounted on the mounting box 1, the main gear is mounted on the output shaft of the drive motor, and the driven gear is mounted on the mounting cylinder 5. The main gear and the driven gear mesh, and at this time, the positioning rod 34 is fixedly connected to the mounting cylinder 5.

[0109] Please see Figure 5 and Figure 6 As a preferred embodiment, the machine vision-based intelligent identification and control device for corn diseases and pests further includes a cleaning component 7. The cleaning component 7 includes a storage box 71, a cleaning brush 72, a rotating shaft 73, and a rotating component. The storage box 71 is mounted on the mounting cylinder 5. One end of the brush handle 721 of the cleaning brush 72 is rotatably mounted on the storage box 71 via the rotating shaft 73. The rotating component is used to drive the cleaning brush 72 to rotate.

[0110] The cleaning component 7 is preferably located on the side adjacent to the detection camera 42. The cleaning angle can be preset. When cleaning is required, if the detection camera 42 is not at the cleaning angle, the drive component 9 drives the mounting cylinder 5 to rotate to the cleaning angle.

[0111] By setting up the cleaning component 7, when the lens of the outdoor detection camera 42 is obstructed by plant stems and leaves (such as when dead plant stems and leaves are blown onto the lens of the detection camera 42 by the wind), the rotating component first drives the cleaning brush 72 to rotate 90 degrees, causing it to rotate out of the storage box 71 and into a vertical position; using the lifting device 3, the height of the detection camera 42 is lowered so that it is aligned with the cleaning brush 72, such as... Figure 6 Then, the drive component 9 drives the mounting cylinder 5 to rotate, thereby driving the detection camera 42 to rotate back and forth, so that the lens of the detection camera 42 and the cleaning cotton 722 of the cleaning brush 72 reciprocate to form a wiping effect, so as to remove foreign objects such as plant stems and leaves attached to the lens.

[0112] Subsequently, the lifting device 3 and the driving component 9 are used to move the detection component 4 back to its original position for continued monitoring.

[0113] The cleaning brush 72 rotates into the storage box 71 with the brush handle 721 facing upwards, thus protecting the cleaning cotton 722 from being exposed to the outdoors and accumulating dust and other impurities, which would affect the cleaning effect.

[0114] Furthermore, the cleaning sponge 722 can be detached and connected to the brush handle 721, making it easy to replace the cleaning sponge 722.

[0115] When the detection camera 42 and the cleaning cotton 722 are at the same height, the cleaning cotton 722 and the detection camera 42 have a significant overlap. Figure 6 This ensures that the cleaning cotton 722 can contact the lens of the detection camera 42 for wiping while rotating.

[0116] The top of the storage box 71 and the end away from the mounting box 1 are open.

[0117] As an optional embodiment, the rotating component includes a rotary motor, which is mounted on the storage box 71. The output shaft of the rotary motor is fixedly connected to the rotating shaft 73, and the rotary motor drives the cleaning cotton 722 to rotate.

[0118] Please see Figure 4 and Figure 5 As another alternative to this example, the machine vision-based intelligent identification and control device for corn diseases and pests also includes a pushing device 6 and a transmission component 8.

[0119] The rotating component is a first gear 74, which is fixed to one end of the rotating shaft 73. The bottom end of the positioning rod 34 is slidably connected to the top of the mounting box 1 through a slider 35.

[0120] The pushing device 6 includes a push cylinder 61 and an L-shaped frame 62. The push cylinder 61 is mounted on the mounting box 1. The output end of the push cylinder 61 passes through the side wall of the mounting box 1 and is connected to the L-shaped frame 62. The top end of the L-shaped frame 62 passes through the top of the mounting box 1 through the sliding hole 101 and is connected to the slider 35.

[0121] The transmission component 8 includes a connecting rod 81 and a first toothed plate 82. One end of the connecting rod 81 passes through the other side wall of the mounting box 1 and is connected to the L-shaped frame 62. The first toothed plate 82 is installed at the other end of the connecting rod 81 and is aligned with the lower side of the first gear 74.

[0122] The driving component 9 includes a sleeve 91, a connecting shaft 92, and a connector 93. The sleeve 91 is fixedly installed in the mounting cylinder 5. One end of the connecting shaft 92 is slidably connected inside the sleeve 91, and the other end is aligned with the insertion hole 211 opened on the drive shaft 21. The connector 93 connects the connecting shaft 92 and the slider 35.

[0123] The opening of the U-shaped sleeve 33 is horizontal and aligned with the strip-shaped opening 501, and is in the opposite direction to the mounting plate 41.

[0124] When drive motor 2 is in lifting mode, such as Figure 3 and Figure 4 At this time, the U-shaped sleeve 33 is fitted on the positioning rod 34, which limits the nut 32 in the rotation direction. When the drive motor 2 drives the lead screw 31 to rotate through the drive shaft 21, the nut 32 moves along the lead screw 31 and the positioning rod 34 to realize the lifting function.

[0125] When the drive motor 2 is in the rotation adjustment state, the push cylinder 61 pulls the slider 35 through the L-shaped frame 62. The slider 35 drives the connecting shaft 92 to be inserted into the insertion hole 211 of the drive shaft 21 through the connector 93. At this time, the slider 35 drives the positioning rod 34 to separate from the U-shaped sleeve 33. The positioning rod 34 moves out of the mounting cylinder 5 from the strip-shaped opening 501. At the same time, the L-shaped frame 62 drives the first toothed plate 82 to act with the first gear 74 through the connecting rod 81, so that the cleaning brush 72 rotates upward by ninety degrees. Figure 6 ;

[0126] At this time, when the drive shaft 21 of the drive motor 2 drives the lead screw 31 to rotate, it drives the mounting cylinder 5 to rotate synchronously through the drive component 9. Since the positioning rod 34 will not be limited by the nut 32, the mounting cylinder 5 drives the mounting plate 41 to rotate through the strip-shaped opening 501. The mounting plate 41 drives the detection camera 42 to rotate, thereby adjusting the angle of the detection camera 42. Image acquisition can be performed on other positions as needed.

[0127] When it is necessary to clean the lens of the inspection camera 42, the inspection camera 42 is rotated back and forth and the cleaning brush 72 is used to wipe it.

[0128] This drives the motor 2 to have a lifting state and a rotation adjustment state, and during the state switching process, the cleaning brush 72 is removed from the storage box 71 for wiping and cleaning the lens of the detection camera 42.

[0129] When the drive motor 2 drives the lead screw 31 to rotate for lifting, the drive motor 2 rotates an integer number of revolutions or an integer and half a revolution each time, so that the socket 211 can be kept aligned with the connecting shaft 92, which facilitates the assembly of the connecting shaft 92 and the socket 211.

[0130] When switching from the rotation adjustment state to the lifting state, the drive motor 2 operates, causing the mounting cylinder 5 to move the detection camera 42 to the desired position. Figure 3 and Figure 4 In this state, the positioning rod 34 can be aligned with the U-shaped sleeve 33 again, which facilitates the reassembly of the positioning rod 34 and the U-shaped sleeve 33. The position angle can be set as the switching angle.

[0131] The mounting box 1 is equipped with a limit rod, and the L-shaped frame 62 is fitted on the limit rod. When the push cylinder 61 pushes or pulls the L-shaped frame 62, the L-shaped frame 62 moves along the limit rod, thereby ensuring the smoothness of the movement.

[0132] The push cylinder 61 is a pneumatic cylinder, a hydraulic cylinder, or an electric push rod.

[0133] A slide rail is provided on the top of the mounting box 1, and the slider 35 slides into the slide rail to form a sliding fit.

[0134] Please see Figure 4 and Figure 5 As an optional embodiment, the connector 93 is a connecting ring, the sleeve 91 has strip holes on both sides, the connecting shaft 92 has a rectangular through hole 921, the connector 93 is sleeved on the outside of the drive shaft 21, one end is fixedly connected to the slider 35, and the other end passes through the strip hole and the rectangular through hole 921.

[0135] When the slider 35 drives the connecting shaft 92 to insert into the socket 211 to a preset depth via the connector 93, the center of the connector 93 coincides with the center of the drive shaft 21.

[0136] When the pushing device 6 moves the slider 35, the slider 35 drives the connecting shaft 92 to be inserted into the insertion hole 211 via the connecting member 93 (connecting ring). When it is inserted to a preset depth, the center of the connecting member 93 coincides with the axis of the drive shaft 21. Figure 7 Therefore, when the drive shaft 21 drives the mounting cylinder 5 to rotate via the connecting shaft 92 and the sleeve 91, the drive shaft 21 can rotate and move along the connecting piece 93 (connecting ring), thereby further improving the stability of the drive shaft 21 during rotation.

[0137] As an optional embodiment, the connector 93 can also be a semi-ring, with one end fixed to the slider 35 and the other end passing through the strip hole and the rectangular through hole 921.

[0138] Please see Figure 4 and Figure 7 A positioning groove 922 is provided on one side of the inner wall of the rectangular through hole 921, and a protrusion 931 is provided on one side of the connector 93 and inserted into the positioning groove 922.

[0139] In this embodiment, the width of the rectangular through hole 921 is greater than the diameter of the connector 93 (connecting ring);

[0140] By setting the protrusion 931 to be inserted into the positioning groove 922, since one end of the connector 93 is fixedly connected to the slider 35, the connecting shaft 92 can be limited in the rotation direction. Thus, the connecting shaft 92, together with the sleeve 91, limits the installation cylinder 5 in the horizontal rotation direction, so that the connecting shaft 92 will not rotate and can remain aligned with the insertion hole 211.

[0141] Among them, such as Figure 7 Middle (a) to Figure 7 In (b), when the slider 35 moves the connector 93, the connector 93 first moves the protrusion 931 out of the positioning groove 922. At this time, the connector 93 (connecting ring) fits against the inner wall of the rectangular through hole 921. As a result, when the slider 35 continues to pull the connector 93, it drives the connecting shaft 92 to be inserted into the insertion hole 211.

[0142] Subsequently, the pusher cylinder 61 pushes the slider 35 through the L-shaped frame 62, which drives the connector 93 to move the connecting shaft 92 out of the insertion hole 211. During this process, the protrusion 931 first inserts into the positioning groove 922, and then pushes the connecting shaft 92 into the sleeve 91.

[0143] Second Embodiment

[0144] Please see Figure 8 Based on the first embodiment, the present invention also proposes a second embodiment. The intelligent identification and control device for corn diseases and pests based on machine vision provided in the second embodiment differs from the first embodiment in that the cleaning brush 72 includes an assembly rod 723, a connecting plate 724, a one-way gear 725, a second toothed plate 83, and a plurality of cleaning sponges 722. The connecting plate 724 is horizontally mounted on the brush handle 721. The assembly rod 723 passes through the connecting plate 724 and is rotatably connected to the connecting plate 724. The plurality of cleaning sponges 722 are evenly mounted on the assembly rod 723 in the circumferential direction. The one-way gear 725 is mounted on the bottom end of the assembly rod 723. The second toothed plate 83 is mounted on the connecting rod 81 and is spaced apart from the first toothed plate 82.

[0145] When the rotating shaft 73 drives the cleaning brush 72 to rotate from horizontal to vertical, the second toothed plate 83 aligns with the one-way gear 725.

[0146] When the push cylinder 61 drives the slider 35 via the L-shaped frame 62 to insert the connecting shaft 92 into the insertion hole 211 to the preset depth, the first toothed plate 82 first meshes with the first gear 74, causing the assembly rod 723 to rotate 90 degrees. The rotating shaft 73 causes the cleaning brush 72 to stand up and rotate out of the storage box 71. After the cleaning brush 72 stands up, the one-way gear 725 aligns with the second toothed plate 83. The connecting rod 81 continues to drive the second toothed plate 83 to move. The second toothed plate 83 meshes with the one-way gear 725, causing the assembly rod 723 to rotate at a preset angle, switching the cleaning cotton 722. This allows for the automatic replacement of the cleaning cotton 722 periodically, avoiding frequent replacements.

[0147] The number of cleaning cotton 722 is no less than three. When there are three, the second toothed plate 83 drives the one-way gear 725 to rotate 120° each time. When there are four, it rotates 90° each time.

[0148] The one-way gear 725 includes a one-way bearing and a gear. The one-way bearing is mounted on the mounting rod 723, and the gear is mounted on the one-way bearing.

[0149] When the connecting shaft 92 separates from the insertion hole 211, and the second toothed plate 83 interacts with the one-way gear 725 again, it will not drive the assembly rod 723 to rotate through the one-way gear 725, that is, it will not switch the cleaning cotton 722.

[0150] Preferably, when Figure 7 In the middle (c), when the connecting shaft 92 is inserted into the insertion hole 211 and the center of the connecting piece 93 is concentric with the drive shaft 21, only the first toothed plate 82 and the first gear 74 act, causing the cleaning brush 72 to rotate upwards vertically. When it is necessary to switch the cleaning cotton 722, the push cylinder 61 continues to pull the connecting rod 81 through the L-shaped frame 62. At this time, the second toothed plate 83 and the one-way gear 725 act to switch different cleaning cotton 722. Then the push cylinder 61 pushes the L-shaped frame 62, so that the connecting piece 93 is concentric with the drive shaft 21 again, and the cleaning work is carried out subsequently.

[0151] The rotating shaft 73 has a fixed anti-slip sleeve on its surface, which increases the friction with the storage box 71. This prevents the cleaning brush 72 from rotating out of the storage box 71 without being subjected to external force, and keeps it in a stable state.

[0152] The cleaning cotton 722 and the assembly rod 723 are detachable.

[0153] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A machine vision-based intelligent identification and control device for corn diseases and pests, characterized in that, include: Image acquisition module, the image acquisition module being used to acquire images of corn plants; An image processing module is used to preprocess the acquired images; the preprocessing includes size normalization, noise reduction, illumination and color correction, image sharpening, pixel normalization, and invalid image filtering. A recognition and analysis module is constructed based on a deep learning model. This module is used to identify and analyze the collected maize plant images and generate prevention and control plans. The execution module is used to implement corresponding prevention and control treatments on corn plants in areas with pests and diseases; The data management module is used to store and manage preprocessed images, pest and disease identification results, and field control operation records.

2. The intelligent identification and control device for corn diseases and pests based on machine vision according to claim 1, characterized in that, The method for constructing the identification and analysis module specifically includes the following steps: S1. Collect images of corn at different growth stages, including seedling stage, jointing stage, and ear stage, including healthy plants and plants with pests and diseases. S2. Label, preprocess, and augment the acquired image data; S3. Divide the processed image data into a dataset, including a training set, a validation set, and a test set; S4. Select a lightweight deep learning object detection model as the basic framework, input the divided training set samples into the lightweight deep learning object detection model, and perform iterative training. S5. Use the test set to test the trained model; S6. Deploy the trained model to the corn disease and pest identification and control system.

3. The intelligent identification and control device for corn diseases and pests based on machine vision according to claim 1, characterized in that, The image acquisition module includes a fixed acquisition unit and a mobile acquisition unit. The fixed acquisition units are distributed in multiple locations in the field. Each fixed acquisition unit includes a mounting box, a lifting device, and a detection component. The lifting device is mounted on the mounting box. The detection component includes a mounting plate and a detection camera. The mounting plate is mounted on the output end of the lifting device, and the detection camera is mounted on the mounting plate via a pitch and rotation component.

4. The intelligent identification and control device for corn diseases and pests based on machine vision according to claim 3, characterized in that, The lifting device includes a drive motor, a lead screw, a nut, a U-shaped sleeve, and a positioning rod. The drive motor is installed inside the mounting box. The bottom end of the lead screw is fixedly connected to the drive shaft of the drive motor. The nut is threaded onto the lead screw. The positioning rod is arranged adjacent to the lead screw. The nut is sleeved on the positioning rod through the U-shaped sleeve. The mounting plate is installed on the nut.

5. The intelligent identification and control device for corn diseases and pests based on machine vision according to claim 4, characterized in that, The top of the mounting box is rotatably mounted with a mounting cylinder, which is sleeved outside the lead screw and positioning rod. The top of the lead screw is rotatably connected to the mounting cylinder. Both sides of the mounting cylinder have slotted openings. One end of the mounting plate passes through one of the slotted openings and is fixedly connected to the nut. The machine vision-based intelligent identification and control device for corn diseases and pests also includes a drive component, which is used to drive the mounting cylinder to rotate.

6. The intelligent identification and control device for corn diseases and pests based on machine vision according to claim 5, characterized in that, It also includes a cleaning component, which includes a storage box, a cleaning brush, a rotating shaft, and a rotating component. The storage box is mounted on the mounting cylinder, one end of the cleaning brush handle is rotatably mounted on the storage box via the rotating shaft, and the rotating component is used to drive the cleaning brush to rotate.

7. The intelligent identification and control device for corn diseases and pests based on machine vision according to claim 6, characterized in that, It also includes a driving mechanism and transmission components; The rotating component is a first gear, which is fixed to one end of the rotating shaft. The bottom end of the positioning rod is slidably connected to the top of the mounting box via a slider. The pushing device includes a push cylinder and an L-shaped frame. The push cylinder is mounted on the mounting box. The output end of the push cylinder passes through the side wall of the mounting box and is connected to the L-shaped frame. The top end of the L-shaped frame passes through the top of the mounting box through a sliding hole and is connected to the slider. The transmission component includes a connecting rod and a first toothed plate. One end of the connecting rod passes through the other side wall of the mounting box and is connected to the L-shaped frame. The first toothed plate is installed at the other end of the connecting rod and is aligned with the lower side of the first gear. The driving component includes a sleeve, a connecting shaft, and a connector. The sleeve is fixedly installed in the mounting cylinder. One end of the connecting shaft is slidably connected inside the sleeve, and the other end is aligned with an insertion hole on the drive shaft. The connector connects the connecting shaft and the slider.

8. The intelligent identification and control device for corn diseases and pests based on machine vision according to claim 7, characterized in that, The connector is a connecting ring. The sleeve has strip-shaped holes on both sides and a rectangular through hole on the connecting shaft. The connector is sleeved on the outside of the drive shaft, with one end fixedly connected to the slider and the other end passing through the strip-shaped hole and the rectangular through hole. When the slider drives the connecting shaft to insert into the socket to a preset depth via the connector, the center of the connector coincides with the center of the drive shaft.

9. The intelligent identification and control device for corn diseases and pests based on machine vision according to claim 8, characterized in that, A positioning groove is provided on one side of the inner wall of the rectangular through hole, and a protrusion is provided on one side of the connector to be inserted into the positioning groove.

10. The intelligent identification and control device for corn diseases and pests based on machine vision according to claim 7, characterized in that, The cleaning brush includes an assembly rod, a connecting plate, a one-way gear, a second toothed plate, and multiple cleaning sponges. The connecting plate is horizontally mounted on the brush handle. The assembly rod passes through the connecting plate and is rotatably connected to the connecting plate. The multiple cleaning sponges are evenly mounted on the assembly rod circumferentially. The one-way gear is mounted at the bottom end of the assembly rod. The second toothed plate is mounted on the connecting rod and is spaced apart from the first toothed plate. When the rotating shaft drives the cleaning brush to rotate from horizontal to vertical, the second toothed plate aligns with the one-way gear.

Images