Machine vision-based directional bird repelling method and system

Abstract

The application discloses a directional bird repelling method and system based on machine vision, and the method comprises the following steps: acquiring an image to be processed; inputting the image to be processed into a trained bird recognition model to output a bird recognition result; if there is a bird, issuing an instruction to a rudder by an industrial computer, adjusting the position of a surface laser emitter by the rudder, and adjusting the emission power of the surface laser emitter by the industrial computer to perform laser sweeping on the bird, so that the bird is repelled. The application reduces the damage of birds to crops, combines image processing technology, machine learning and light control, and realizes accurate and harmless bird repelling.

Description

Technical Field

[0001] This invention relates to the field of bird deterrence technology, and in particular to a method and system for directional bird deterrence based on machine vision. Background Technology

[0002] Birds may invade human territory in places such as farms, orchards, wind farms, and military and civilian airports, endangering people's work or the safety of equipment. Especially at airports, in order to protect the safety of aircraft take-off and landing and prevent birds from being sucked into aircraft engines or colliding with the aircraft fuselage, landing gear, tail, windshield, and all other parts of the aircraft, all measures must be taken to prevent bird intrusion.

[0003] In addition, when birds are driven away, they may fly away from the driver in a random direction due to their stress response. This randomness and uncontrollability may cause birds to crash into other important devices when they are driven away. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a machine vision-based method and system for directional bird deterrence, reducing bird damage to crops. This system combines image processing technology, machine learning, and lighting control to achieve precise and harmless bird deterrence.

[0005] On the one hand, a machine vision-based method for directional bird deterrence is provided, including:

[0006] Obtain the image to be processed;

[0007] The image to be processed is input into the trained bird recognition model, and the bird recognition result is output.

[0008] If birds are present, the industrial control computer sends a command to the servo motor, which controls the laser emitter on the control surface to adjust its position. The industrial control computer also controls the laser emitter's emission power to perform a laser scan on the birds, thereby driving them away.

[0009] On the other hand, a machine vision-based directional bird deterrence system is provided, including: a surface laser emitter;

[0010] The surface laser emitter is mounted on a servo motor, which is mounted on a base. The servo motor is connected to an industrial control computer, and the surface laser emitter is also connected to the industrial control computer. The industrial control computer is also connected to a camera, a photosensor, a communication module, and a touch screen.

[0011] If the industrial control computer detects birds, it sends a command to the servo motor. The servo motor then controls the laser emitter on the control surface to adjust its position and the industrial control computer controls the laser emitter's emission power to perform a laser scan on the birds, thereby driving them away.

[0012] The above technical solution has the following advantages or beneficial effects:

[0013] This invention combines YOLOv9 network image processing technology, machine learning, and lighting control to achieve precise and harmless bird deterrence. When birds are directed upwards, the servo rotates the surface laser emitter to a horizontal position, sweeping upwards during the deterrence; when birds are directed to the right, the servo rotates the surface laser emitter to a vertical position, sweeping from left to right during the deterrence. Attached Figure Description

[0014] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0015] Figure 1 This is a schematic diagram of the connection relationship in this invention.

[0016] Figure 2 This is a front view of the surface laser emitter of the present invention;

[0017] Figure 3 This is a side view of the surface laser emitter of the present invention. Detailed Implementation

[0018] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] Example 1

[0020] This embodiment provides a machine vision-based method for directional bird deterrence;

[0021] like Figure 1 As shown, the machine vision-based directional bird deterrence method includes:

[0022] S101: Obtain the image to be processed;

[0023] S102: Input the image to be processed into the trained bird recognition model and output the bird recognition result;

[0024] S103: If birds are present, the industrial control computer sends a command to the servo motor, which controls the laser emitter on the control surface to adjust its position, and the industrial control computer controls the laser emitter's emission power to perform laser scanning on the birds, thereby driving them away.

[0025] Furthermore, the acquisition of the image to be processed is achieved by capturing images of the target area using a camera.

[0026] Furthermore, the trained bird recognition model is implemented using a YOLOv9 network.

[0027] Furthermore, the training process of the trained bird recognition model includes:

[0028] Construct training and testing sets; both training and testing sets include: monitoring images of known bird outline regions;

[0029] The training set is input into the bird recognition model, the surveillance image is used as the input value of the model, and the bird outline region is used as the output value of the model. When the total loss function value of the model no longer decreases, or when the number of iterations reaches the set number, the training is stopped, and the preliminary trained bird recognition model is obtained.

[0030] The test set is input into the initially trained bird recognition model to test the model. If the test accuracy is higher than the set threshold, it means that the current model is the final trained bird recognition model. If it is lower than the set threshold, the training set is changed and the model is retrained.

[0031] Furthermore, if birds are present, the industrial control computer sends a command to the servo motor, which controls the laser emitter on the control surface to adjust its position. The industrial control computer also controls the emission power of the laser emitter to perform a laser scan to drive away the birds. Specifically, this includes:

[0032] If the birds to be driven away are located above or below the center point of the surface laser emitter, the industrial control computer sends a command to the servo motor, which adjusts the surface laser emitter so that its central axis is parallel to the horizontal line; and controls the laser emitted by the surface laser emitter to scan from top to bottom.

[0033] If the birds to be driven away are located to the left or right of the center point of the surface laser emitter, the industrial control computer sends a command to the servo motor, which adjusts the surface laser emitter so that its central axis is perpendicular to the horizontal line; and controls the laser emitted by the surface laser emitter to scan from left to right.

[0034] For example, when the bird is driven upwards, the servo rotates the surface laser emitter to a horizontal position and performs a sweeping motion from bottom to top during the driving motion; when the bird is driven to the right, the servo rotates the surface laser emitter to a vertical position and performs a sweeping motion from left to right during the driving motion.

[0035] Furthermore, if birds are present, the industrial control computer sends a command to the servo motor, which controls the laser emitter on the control surface to adjust its position. The industrial control computer also controls the emission power of the laser emitter to perform a laser scan to drive away the birds. Specifically, this includes:

[0036] The industrial control computer collects the intensity of ambient light through a photosensitive sensor. If the intensity of ambient light is higher than the first set threshold, the emission power of the laser emitter on the control surface of the industrial control computer is greater than the second set threshold, and then the laser is used to sweep the birds to drive them away.

[0037] The industrial control computer collects the intensity of ambient light through a photosensitive sensor. If the intensity of ambient light is less than the third set threshold, the emission power of the laser emitter on the control surface of the industrial control computer is greater than the fourth set threshold but less than the second set threshold. Then, the laser is used to scan the birds and drive them away.

[0038] Furthermore, the method also includes:

[0039] When two bird targets repeatedly appear within a set time range and a set location area, the two target areas are compared. If the similarity between the two target areas is higher than a set threshold, it indicates that the current bird target is interference information and will not be driven away again.

[0040] Furthermore, the surface laser emitter is mounted on a servo motor, which is mounted on a base. The servo motor is connected to an industrial control computer, and the surface laser emitter is also connected to the industrial control computer. The industrial control computer is also connected to a camera, a photosensor, a communication module, and a touch screen.

[0041] Furthermore, the photosensitive sensor is installed on the side and rear of the camera, and controls the laser emission intensity of the surface laser emitter according to the intensity of natural light. When the intensity of natural light is higher than a set threshold, the laser intensity of the surface laser emitter is increased, and vice versa.

[0042] Furthermore, the touch display screen is used to visualize various situations and enable manual operation.

[0043] This invention provides a machine vision-based directional bird deterrence system, comprising an industrial control computer, a bird monitoring device for real-time monitoring of the presence of birds in the surrounding environment, a surface laser emitter for deterring birds, and a power supply. It uses the YOLOv9 deep learning model for target tracking and monitoring. Upon detecting a bird target, the surface laser emitter drives the bird in a specific direction, and the system self-corrects for potential false alarms in the deep learning framework. This invention combines image processing technology, machine learning, and light processing to provide a stable and reliable system that can drive birds in a specific direction, thereby protecting crops, building structures, and ensuring flight safety.

[0044] This system uses a deep learning model to monitor birds, a machine learning method based on multi-layer neural networks to learn data features and patterns. Using the YOLOv9 model and a training set of bird images, it improves and optimizes recognition accuracy, performs inference and decision-making, and drives a surface laser emitter to adjust its posture and emission via a control unit.

[0045] To address the false positives (misclassifications of targets as positive) inherent in deep learning frameworks, a list is created to store targets that appear repeatedly. Often, the same target remains in the same position after being repeatedly removed, indicating a high probability of a false positive. To solve this problem, this invention proposes a method using Interchange of Units (IOU) comparison to determine if a target is the same and adding it to a blacklist.

[0046] The specific scheme for the IOU comparison method is as follows:

[0047] First, a threshold is set as the confidence level for considering two targets as the same target during comparison. Using the (x1, y1, x2, y2) coordinates returned by the model, candidate bounding boxes are defined, and IOU comparisons are performed on similar candidate bounding boxes. The IOU calculation formula is:

[0048]

[0049] Where A represents the area of ​​the first target; B represents the area of ​​the second target;

[0050] When the Interchange of Union (IOU) exceeds a set threshold, the target is considered to be the same. If a consecutive target appears N times, it is added to the false positive list, and the model will no longer react to it.

[0051] However, to prevent missed detections, the model of this invention will clear the false positive list once after every M (M>N) times.

[0052] This example provides a machine vision-based directional bird deterrence system, such as... Figure 2 , Figure 3 As shown, it includes an industrial control computer, a bird detection camera for real-time monitoring of whether birds are approaching, a surface laser emitter for directional bird repulsion, a photosensitive sensor for monitoring the intensity of natural ambient light, a communication module for connecting with other expansion modules, and a human-computer interaction module for convenient user monitoring and customization.

[0053] Bird monitoring cameras use high-resolution cameras to capture images of the surrounding environment and connect to an industrial control computer via communication signals to transmit data.

[0054] The industrial control computer stores a machine learning model, specifically a bird detection model, based on the YOLO (You Only Look Once) v9 deep learning framework. YOLO is renowned for its highly parallel design, allowing the entire image to be processed in a single forward pass. This results in lower inference time, enabling the system to detect birds quickly and effectively in real-time environments. The model's detection accuracy can be further improved using a training dataset. After reasoning and making decisions about the current situation, the bird detection model formulates an intelligent bird deterrence strategy, adjusting the intensity and timing of the laser beam. The control unit then controls the peripheral area laser emitter. Upon receiving communication signals from the industrial control computer, the area laser emitter adjusts its orientation, performing left-right or up-down sweeps to precisely target and drive away the birds.

[0055] The human-computer interaction system is used for system debugging, manual adjustment of various system parameters, and manual operation. Manually adjustable parameters include model selection, light intensity, sweep speed, drive delay event, drive accuracy, etc. The surface laser emitter can also be manually operated for sweeping. It can also be used to debug other expansion modules.

[0056] Example 2

[0057] This embodiment provides a machine vision-based directional bird deterrence system, including: a surface laser emitter;

[0058] The surface laser emitter is mounted on a servo motor, which is mounted on a base. The servo motor is connected to an industrial control computer, and the surface laser emitter is also connected to the industrial control computer. The industrial control computer is also connected to a camera, a photosensor, a communication module, and a touch screen.

[0059] If the industrial control computer detects birds, it sends a command to the servo motor. The servo motor then controls the laser emitter on the control surface to adjust its position and the industrial control computer controls the laser emitter's emission power to perform a laser scan on the birds, thereby driving them away.

[0060] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., 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 machine vision-based method for directional bird deterrence, characterized by: include: Obtain the image to be processed; The image to be processed is input into the trained bird recognition model, and the bird recognition result is output. If birds are present, the industrial control computer sends a command to the servo motor, which controls the laser emitter on the control surface to adjust its position. The industrial control computer also controls the emission power of the laser emitter to perform a laser scan on the birds, thereby driving them away. The method further includes: When two bird targets repeatedly appear in a set location area within a set time range, the two target areas are compared. If the similarity between the two target areas is higher than a set threshold, it means that the current bird target is interference information and will not be driven away again. By creating a list to store targets that appear consecutively, it's possible that the same target remains in the same position after being repeatedly removed, which is likely a false positive by the model. An Interchange of Units (IoU) comparison is used to determine if they are the same target and adds them to a blacklist. The specific IoU comparison method is as follows: First, a threshold is set as the confidence level for considering two targets as the same target during comparison; using the (x1, y1, x2, y2) coordinates returned by the model, candidate bounding boxes are defined, and IoU comparisons are performed on similar candidate bounding boxes; the IoU calculation formula is: in, This indicates the area representing the first target. The region represents the second target; when the IOU is greater than a set threshold, it is considered to be the same target; when a repeated consecutive target appears N times, it is added to the misjudgment list and the model will no longer react to it.

2. The machine vision-based directional bird deterrence method as described in claim 1, characterized in that, The training process for the trained bird recognition model includes: Construct training and testing sets; both training and testing sets include: monitoring images of known bird outline regions; The training set is input into the bird recognition model, the surveillance image is used as the input value of the model, and the bird outline region is used as the output value of the model. When the total loss function value of the model no longer decreases, or when the number of iterations reaches the set number, the training is stopped, and the preliminary trained bird recognition model is obtained. The test set is input into the initially trained bird recognition model to test the model. If the test accuracy is higher than the set threshold, it means that the current model is the final trained bird recognition model. If it is lower than the set threshold, the training set is changed and the model is retrained.

3. The machine vision-based directional bird deterrence method as described in claim 1, characterized in that, If birds are present, the industrial control computer sends a command to the servo motor, which then controls the laser emitter on the control surface to adjust its position. The industrial control computer also controls the laser emitter's emission power to perform a laser scan to drive away the birds. Specifically, this includes: If the birds to be driven away are located above or below the center point of the surface laser emitter, the industrial control computer sends a command to the servo motor, which adjusts the surface laser emitter so that its central axis is parallel to the horizontal line; and controls the laser emitted by the surface laser emitter to scan from top to bottom. If the birds to be driven away are located to the left or right of the center point of the surface laser emitter, the industrial control computer sends a command to the servo motor, which adjusts the surface laser emitter so that its central axis is perpendicular to the horizontal line; and controls the laser emitted by the surface laser emitter to scan from left to right.

4. The machine vision-based directional bird deterrence method as described in claim 1, characterized in that, If birds are present, the industrial control computer sends a command to the servo motor, which then controls the laser emitter on the control surface to adjust its position. The industrial control computer also controls the laser emitter's emission power to perform a laser scan to drive away the birds. Specifically, this includes: The industrial control computer collects the intensity of ambient light through a photosensitive sensor. If the intensity of ambient light is higher than the first set threshold, the emission power of the laser emitter on the control surface of the industrial control computer is greater than the second set threshold, and then the laser is used to sweep the birds to drive them away. The industrial control computer collects the intensity of ambient light through a photosensitive sensor. If the intensity of ambient light is less than the third set threshold, the emission power of the laser emitter on the control surface of the industrial control computer is greater than the fourth set threshold but less than the second set threshold. Then, the laser is used to scan the birds and drive them away.

5. The machine vision-based directional bird deterrence method as described in claim 1, characterized in that, The surface laser emitter is mounted on a servo motor, which is mounted on a base. The servo motor is connected to an industrial control computer, and the surface laser emitter is also connected to the industrial control computer. The industrial control computer is also connected to a camera, a photosensor, a communication module, and a touch screen.

6. The machine vision-based directional bird deterrence method as described in claim 5, characterized in that, The photosensitive sensor is installed on the side and rear of the camera. It controls the laser emission intensity of the surface laser emitter according to the intensity of natural light. When the intensity of natural light is higher than a set threshold, the laser intensity of the surface laser emitter is increased, and vice versa.

7. The machine vision-based directional bird deterrence method as described in claim 1, characterized in that, The acquisition of the image to be processed is achieved by capturing images of the target area using a camera; the trained bird recognition model is implemented using the YOLOv9 network.

8. A machine vision-based directional bird deterrence system, characterized in that, The method for directional bird deterrence based on machine vision as described in any one of claims 1-7 includes: a surface laser emitter; The surface laser emitter is mounted on a servo motor, which is mounted on a base. The servo motor is connected to an industrial control computer, and the surface laser emitter is also connected to the industrial control computer. The industrial control computer is also connected to a camera, a photosensor, a communication module, and a touch screen. If the industrial control computer detects birds, it sends a command to the servo motor. The servo motor then controls the laser emitter on the control surface to adjust its position and the industrial control computer controls the laser emitter's emission power to perform a laser scan on the birds, thereby driving them away.

Images